An attacker has drained nearly 200,000 XRP — worth approximately $200,000 at the time of the incident — from a cross-chain bridge connecting the XRP Ledger to the tx blockchain. The exploit took advantage of a software flaw that made the bridge believe fake deposits were real funds. According to the project team, the attacker created unbacked XRP on the tx chain and then exchanged that token for genuine XRP held in the bridge's reserve wallet. The bridge has since been halted, and the operator has filed a complaint with the FBI.

Cross-chain bridges are a crucial part of the cryptocurrency ecosystem. They allow users to move digital assets from one blockchain to another without going through a centralized exchange. In a typical bridge setup, a user deposits an asset on the source chain, and the bridge locks that asset in a smart contract or a reserve wallet. The bridge then mints an equivalent amount of a token on the destination chain. When the user wants to return to the original chain, the destination token is burned or locked, and the original asset is released. This process depends on the bridge's ability to accurately verify that deposits actually took place.

The XRP Ledger to tx bridge was designed to enable transfer of XRP across these two networks. The bridge holds real XRP in reserve to back the bridged XRP tokens issued on tx. If the bridge issues tokens without a corresponding deposit, those tokens are unbacked. The value of the reserve is then diluted, and users holding bridged tokens can suffer losses. That is exactly what happened in this exploit.

How the Exploit Worked

The attacker identified a flaw in the bridge's software that allowed them to trick the system into recognizing deposits that never occurred. On the tx blockchain, the attacker generated XRP tokens without depositing any actual XRP on the XRP Ledger. The bridge's software logged these as legitimate cross-chain deposits and issued unbacked bridged XRP to the attacker. The attacker then redeemed the unbacked tokens for real XRP from the bridge's reserve wallet.

This type of attack is sometimes called a "fake deposit" exploit. It can happen when the bridge's verification mechanism relies on data that can be forged, or when the software does not properly cross-check the source chain's records. In some cases, a bridge may trust a third-party oracle or relayer to report deposits. If that relayer is compromised or the verification logic has a bug, fake deposits can pass through.

The amount stolen in this incident is relatively small compared to some of the largest bridge hacks in crypto history. However, the event raises concerns about the safety of cross-chain infrastructure, especially for users who rely on bridges to move assets between networks. The fact that the bridge operator had to halt operations after the exploit shows how quickly trust can be eroded.

Response and Current Status

After detecting the exploit, the tx team said it moved quickly to stop further losses. The bridge was halted, and developers deployed a patch to fix the vulnerability. The team also hired blockchain forensics specialists to trace the stolen funds. Additionally, an FBI complaint has been filed, which could help in the investigation and potential recovery of the assets.

Despite these steps, the operator has not yet provided a clear explanation of how affected holders will be compensated. Users who had bridged XRP on tx may be wondering whether their tokens are still backed by real XRP. If the reserve is not restored, those users could face a shortfall. It is also unclear whether any legal recourse is available to those who lost funds as a result of the exploit.

Why Cross-Chain Bridges Are Targeted

Bridges are prime targets for hackers because they often hold large amounts of locked value. When a bridge is compromised, an attacker can gain access to assets that back bridged tokens on multiple chains. The complexity of bridge implementations also makes them vulnerable to bugs. Many bridges use multi-party computation, threshold signatures, or optimistic validation mechanisms, all of which introduce potential attack surfaces.

The XRP Ledger to tx bridge is not alone in facing such issues. Numerous bridge exploits have occurred over the past few years. In 2022, the Ronin bridge was drained of more than $600 million, and the Harmony Horizon bridge lost around $100 million. In 2021, the Poly Network was exploited for over $600 million, although most funds were eventually returned. These incidents highlight repeatedly that cross-chain bridge security is still an unresolved challenge.

Fake deposit vulnerabilities have also been seen in other projects. Some bridges failed to properly validate transaction proofs, allowing attackers to mint tokens without depositing collateral. Others fell victim to "double spend" issues or flaws in smart contract permission logic. Each incident provides a lesson, but the broader ecosystem remains vulnerable because every bridge has its own codebase and trust assumptions.

Impact on the XRP Ecosystem

The exploit is likely to have ripple effects across the XRP ecosystem. Bridged assets are commonly used in decentralized finance applications, and a loss in a bridge's reserve can reduce the liquidity available to users. The tx bridge had been intended to provide interoperability between XRP Ledger and the emerging tx network, which aims to offer lower fees and faster transactions. Now, confidence in that interoperability has been shaken.

For XRP Ledger users, the incident serves as a reminder that cross-chain activity carries risks beyond those of the base network. Even though the XRP Ledger itself was not compromised, the bridge's connection to it became a point of failure. This is a common pattern: the underlying blockchain is secure, but the bridge software acting as an intermediary is not.

The timing also matters. As institutional interest in crypto grows, security incidents like this can influence regulatory scrutiny. Law enforcement involvement via the FBI complaint may lead to broader investigations into bridge operators and their compliance practices. It may also encourage other bridge projects to review their own verification logic before falling victim to a similar attack.

Security Lessons and Future Outlook

One clear lesson from this incident is that bridges must treat deposit verification as a critical security function. Relying on software to confirm deposits across chains requires robust validation mechanisms. This can include checking the consensus or finality of the source chain, using multiple independent verifiers, and implementing bug bounty programs to catch vulnerabilities before attackers do.

Another lesson is the importance of transparency after an attack. The tx team has communicated some details about the exploit and its response, but it has not yet offered a full postmortem or a compensation plan. Clear communication is essential to maintaining user trust and preventing market panic. In previous bridge hacks, some projects were able to restore funds or reimburse users, while others failed and eventually shut down.

For security researchers, this exploit adds to a growing catalogue of bridge vulnerabilities. It may prompt further audits of similar bridging systems, especially those that rely on software to interpret deposit data from another chain. The use of blockchain forensics is also becoming standard practice after major thefts, as teams work to trace funds and cooperate with authorities.

At a broader level, the incident underscores the tension between innovation and security in decentralized finance. Bridges are necessary for an interconnected multi-chain world, but they also create concentration risks. If one bridge fails, everyone who depends on it can be affected. Developers must therefore design bridges with fail-safes, limits on transfer amounts, and emergency pause mechanisms to mitigate the impact of future exploits.

What Happens Next

The tx bridge remains halted as the team completes its security patch and investigates the attack. Users will likely need to wait for the operator to decide whether to resume operations and how to handle the shortfall. The FBI complaint indicates that law enforcement is now involved, which may complicate but also strengthen the recovery effort. Blockchain forensics experts may be able to track the stolen XRP to exchange wallets, but there is no guarantee that the funds will be returned.

In the meantime, bridge users are advised to remain cautious. Checking a bridge's security history, audit reports, and emergency plans before depositing funds is essential. The crypto industry has seen too many bridge failures to ignore the risks. As for the XRP Ledger to tx bridge, the future is uncertain. What is clear, however, is that a small software mistake led to a costly exploit, and the consequences will be felt by all those who trusted the bridge with their assets.

Source: Coindesk News

A rollback would mean discarding the blocks created after a specific point and rebuilding the chain from an earlier state. For users, that could reverse payments, deposits and withdrawals that occurred after the first invalid block was mined on Aug. 7. For exchanges and payment processors, it would create a backlog of reconciliation tasks and potentially leave some transactions orphaned. The decision is not yet final, and the network may still choose to preserve the current chain if enough nodes reject the replacement.

What went wrong?

Ravencoin’s code is based on Bitcoin Core, but it uses a different mining algorithm called KawPow, which is designed to give ordinary GPU miners a chance to participate. Bitcoin itself is protected by a set of consensus rules that every node checks before accepting a new block. If a block violates those rules, it is simply rejected. On Ravencoin, however, a bug in the software allowed a small number of invalid blocks to be propagated across the network. The attack was first detected when nodes began accepting blocks that should have been refused under the protocol’s own consensus rules.

The identity of the attacker is not known. Some community members believe the flaw was exploited deliberately to test the network or to profit from the resulting chaos. Others think the invalid blocks might have been produced accidentally by a miner running outdated or modified software. Regardless of intent, the effect was the same: the chain’s integrity was called into question, and confidence in Ravencoin’s reliability was shaken.

Why do mining pools control the outcome?

In proof-of-work blockchains, validators are not chosen by stake but by computing power. The longest chain with the most accumulated proof of work is generally considered the valid chain. Because 2Miners and RavenMiner control the majority of Ravencoin’s hashrate, they can choose to build on top of an earlier point in the chain and ignore the invalid blocks. If other miners and node operators follow them, the network will effectively abandon the blocks that were added after the flaw was exploited.

This is not the first time a cryptocurrency has faced a chain reorganization, but it is unusual for a modern asset. Bitcoin experienced some large reorganizations in its early years, mostly due to bugs or intentional attacks that were later resolved. Ethereum also made a deliberate - and highly controversial - hard fork in 2016 to reverse the effects of a hack on The DAO. In that case, the community decided that preventing losses was more important than preserving immutability. Ravencoin’s situation is similar in spirit, though the stakes are smaller.

What happens to transactions?

If the rollback proceeds, any transaction that was confirmed after the first invalid block could be removed from the ledger. That includes deposits made to exchanges, withdrawals sent to wallets, and payments between users. The original balances at the time of the fork would be restored. This is technically possible because the replacement chain will not contain those transactions. The mining pools behind the rollback are essentially saying that the post-Aug. 7 state of the network is invalid and should be forgotten.

For ordinary users, the immediate risk is that they might see a transaction reversed hours after receiving a confirmation. If they spent the received funds or transferred them to another address, they could end up owing money. Merchants that accepted some of the affected payments would be particularly vulnerable. This kind of uncertainty is why exchanges have reacted so quickly.

Exchanges move to protect users

Several trading platforms have announced a halt to Ravencoin deposits and withdrawals. European exchange Bitvavo stopped RVN processing to prevent users from depositing coins that might later be rolled back. Upbit, a major exchange in South Korea, also paused its RVN services. These measures are preventive. If a deposit is made and then reversed on the chain, an exchange could credit the user for coins that effectively disappear, leading to losses for the platform. By pausing activity, exchanges are waiting to see which chain becomes canonical.

The halts could also delay legitimate trading. Users who want to sell RVN may not be able to move their coins to an exchange, and some may find that their balances are temporarily frozen. The exchanges are caught in a difficult position: if they resume trading before the rollback is fully resolved, they risk accepting orphaned transactions; if they wait too long, they lose customer trust.

Market response

The uncertainty has taken a toll on Ravencoin’s price. RVN, the network’s native token, has fallen sharply since news of the flaw spread. The decline reflects a broader loss of confidence in a project that was already competing with larger smart-contract platforms. Bitcoin’s market dominance and the rise of alternative token standards on Ethereum have made it harder for smaller networks like Ravencoin to attract attention. A rollback, even if successful, will not erase the reputational damage from this event.

Long-term holders may see this as a test of Ravencoin’s governance. The project does not have a formal treasury or a strong development foundation like some other cryptocurrencies. It relies on community coordination and voluntary contributions. The ability to organize a rollback through mining pools shows that the network can act decisively when necessary, but it also reveals how few actors have real power. A majority of hashrate is effectively the final authority, and that is a fact that cannot be ignored.

What is Ravencoin?

Ravencoin was launched in January 2018 as an open-source project tied to the Bitcoin codebase. It was created by Bruce Fenton and Joel Weight, with the goal of building a platform specifically for token issuance and peer-to-peer asset transfers. Unlike Ethereum, which supports complex smart contracts, Ravencoin’s protocol is intentionally simpler. Users can create tokens that have fixed supply, limited supply, or special privileges such as mining rewards. The project’s native asset is RVN, which is used for transaction fees and as a vehicle for transferring value across the network.

The network has a loyal following among GPU miners because KawPow uses less memory than some other algorithms and is easier for home miners to run. However, Ravencoin remains a relatively small ecosystem. Its market capitalization is a fraction of Bitcoin’s, and its usage is mostly limited to token experiments and speculative trading. The current crisis could push it to improve its software and testing processes, but it could also drive users away.

Can the rollback be prevented?

Not everyone in the Ravencoin community agrees that a rollback is the right response. Some argue that even invalid blocks are part of history and that altering the chain would set a dangerous precedent. They believe that the network should accept the current chain, fix the bug, and move on. Others are worried that preserving the flawed chain would allow invalid tokens or transactions to remain, which could cause further problems down the line.

The outcome depends on node operators. Mining pools can produce replacement blocks, but they cannot force other nodes to accept their chain. If a significant number of node operators refuse to switch, the network could split into two competing branches. This is called a hard fork. Both chains would share the same history up to the fork point, but they would diverge afterwards. The value of RVN would then be divided between the two chains, causing additional confusion.

For now, the two largest mining pools appear to have enough support to move forward. The project’s code repositories have been updated with a patch, and developers are urging miners to update their software. The window for a clean rollback is limited. As new blocks are mined on the original chain, the cost of discarding them increases. The longer the community waits, the harder it will be to coordinate a replacement.

The broader lesson for crypto

Ravencoin’s predicament highlights a fundamental tension in blockchain technology. Immutability is often called one of the greatest value propositions of decentralized ledgers, but it is not automatic. It depends on consensus. When the majority of validators decide to reorganize the chain, the immutable record becomes mutable. This is rare, but it happens. The cryptocurrency industry has seen rollbacks and hard forks before, and it will likely see them again.

The impact on Ravencoin’s ecosystem will depend on how quickly the network can restore normal operations. If the rollback succeeds without major issues, the project might emerge with a stronger software team and clearer decision-making processes. If it fails, the network could be left with a permanent split and a damaged reputation.

For exchanges and users, the immediate priority is to stay informed and avoid sending funds until the network reaches a clear consensus. Deposits and withdrawals are likely to remain suspended for at least a few more days. Traders should also be cautious about speculative positions in RVN, because the price could continue to swing as the community debates the best path forward.

The entire episode is a reminder that no blockchain is completely immune to failure. Even a project built on Bitcoin’s well-tested code can introduce vulnerabilities by modifying the consensus rules. Whether Ravencoin survives this test is still open to question. But the process of trying to undo four days of transactions may become a case study for future networks facing similar threats.

Source: Coindesk News

CoreWeave Beats Estimates, Backlog Tops $100 Billion

CoreWeave's quarterly results underlined the continued strength of AI infrastructure spending. Revenue more than doubled to $2.58 billion from $1.21 billion a year earlier, slightly ahead of the $2.56 billion analysts had expected. The company's revenue backlog, a measure of committed future business, reached roughly $104 billion at the end of June. CoreWeave said that figure does not include more than $25 billion of new customer commitments added early in the third quarter, signaling that demand remains robust.

However, the company's costs are also rising quickly. Operating expenses more than doubled to $2.62 billion, and CoreWeave posted a net loss of $626 million. The company has been racing to add infrastructure, expanding active power by nearly 500 megawatts to 1.5 gigawatts during the quarter, with contracted power reaching about 3.7 gigawatts. Shares jumped 9% in after-hours trading as investors focused on the revenue beat and the growing backlog.

Stock Market Looks to Inflation Data

U.S. stocks ended the day modestly lower as investors positioned ahead of Wednesday's consumer price index report. The broad-market S&P 500 declined a little over 0.3%, while the tech-heavy Nasdaq Composite fell 0.6%. Gold also reversed course from a two-month high above $4,400 an ounce, turning early gains into a 0.5% loss. In the cryptocurrency market, ether, XRP and Solana were largely flat, while bitcoin drifted lower.

Traders are now looking to the U.S. inflation report for July, which could offer fresh clues on the Federal Reserve's next move. Economists expect headline CPI to rise 0.1% from June and 3.4% from a year earlier. Core CPI, which strips out food and energy, is forecast to increase 0.2% on the month and 2.5% on a yearly basis.

Bitcoin Volatility Could Be Awakened

Bitcoin's summer l

Source: Coindesk News

The six partners, which include some of the largest financial institutions on Wall Street, will work with Nvidia to establish what the company calls bankable infrastructure. Under the new framework, AI compute—largely powered by Nvidia’s GPUs in specialized facilities known as AI factories—would be treated as long-lived, revenue-generating assets rather than short-lived technology expenses. That distinction is crucial because it opens the door to project financing, securitization, and other financial instruments typically reserved for traditional infrastructure such as pipelines, data centers, and power plants.

A New Asset Class Takes Shape

Nvidia’s push represents a strategic effort to reframe how investors and corporations view AI computing. Historically, GPUs were considered capital expenditures that depreciated quickly, often over three to five years. But with the rapid growth of AI workloads, the demand for high-performance computing has become more sustained and predictable. Nvidia aims to capitalize on this trend by arguing that AI compute infrastructure, when deployed in AI factories, can generate stable cash flows over ten years or more.

The company has been aggressively expanding its data center business, which has become its largest revenue segment. In recent quarters, Nvidia’s data center revenue has surged as hyperscale cloud providers, enterprises, and governments race to build out AI capabilities. The new financing platforms could accelerate this growth by lowering the cost of capital for AI infrastructure projects and attracting institutional investors who are looking for long-term, yield-generating assets.

According to people familiar with the memorandums, the Wall Street firms will help structure investment vehicles that bundle AI compute capacity into tradable securities. This would allow insurance companies, pension funds, and sovereign wealth funds to gain exposure to AI infrastructure without directly owning and operating data centers. Nvidia’s role would be to provide the GPUs and software stack, while its financial partners would handle the capital markets engineering.

Why This Matters for Crypto Compute

The implications for the crypto industry are profound. Over the past few years, a number of decentralized compute networks have emerged with the goal of pooling idle GPUs from individual miners and data centers to offer a cheaper, more distributed alternative to centralized cloud providers. These networks, often referred to as decentralized physical infrastructure networks (DePIN), allow users to rent computing power for AI training, rendering, and other heavy workloads. Projects like Render Network, Akash Network, and others have tried to position themselves as the Airbnb of GPU compute.

However, Nvidia’s move could widen the gap between these decentralized networks and the massive, capital-intensive AI factories being built by tech giants and well-funded startups. Decentralized compute networks typically rely on consumer-grade GPUs and smaller data centers that are geographically dispersed. While this model offers privacy and censorship resistance, it also struggles with latency, reliability, and scalability. For large-scale AI training runs that require thousands of GPUs working in parallel, the decentralized approach often falls short.

Moreover, the financial engineering that Nvidia is pursuing could make centralized AI compute even more attractive to institutional customers. If AI compute becomes a bankable asset class, companies will find it easier to lease capacity, finance expansions, and secure long-term contracts. This would give centralized providers like Nvidia’s ecosystem a significant advantage in terms of cost efficiency and operational stability.

Nvidia’s Rocky History With Crypto

Nvidia’s relationship with cryptocurrency has been complicated. During the crypto mining boom of 2017 and 2021, the company’s GPUs were in high demand for mining Ethereum and other proof-of-work coins. Nvidia even introduced a line of crypto-specific processors called CMP (Cryptocurrency Mining Processor) to address the shortage of gaming GPUs. However, the company also faced backlash from gamers who struggled to find graphics cards at reasonable prices.

When Ethereum transitioned to proof-of-stake in 2022, the crypto mining demand for GPUs plummeted, and Nvidia redirected its focus firmly toward AI. This pivot proved timely, as the launch of generative AI tools like ChatGPT triggered an explosion in demand for AI compute. Nvidia’s A100 and H100 GPUs became the gold standard for AI training, and the company’s market capitalization soared past $2 trillion.

Despite this pivot, Nvidia has not entirely abandoned the crypto sector. The company has expressed interest in zero-knowledge proof acceleration and other cryptographic workloads that require parallel processing. But the scale of investment in AI infrastructure dwarfs anything seen in crypto. The $500 billion in potential financing is orders of magnitude larger than the total market capitalization of all decentralized compute tokens combined.

Wall Street’s Role in the AI Arms Race

The involvement of Wall Street is a clear signal that AI infrastructure is being treated as a mainstream asset class. The six firms that signed the memorandums have not been publicly named, but reports suggest they include some of the biggest banks and asset managers in the United States. Their participation could lead to the creation of publicly listed funds and exchange-traded products that give retail investors exposure to AI compute.

This development mirrors what happened in the crypto mining industry a few years ago, when Wall Street began financing large-scale Bitcoin mining facilities. Companies like Marathon Digital and Riot Platforms went public and raised billions of dollars to build massive mining farms. However, the AI compute market is significantly larger and more diverse, encompassing everything from hyperscale cloud providers to enterprise edge deployments.

By turning AI compute into a bankable infrastructure asset, Nvidia is also insulating itself from chip cycle volatility. If a downturn in AI spending occurs, the company can rely on long-term contracts and financing partnerships to smooth out revenue. This is a smart strategic move that locks customers into multi-year commitments and makes it harder for competitors like AMD or Intel to gain traction.

Decentralized Networks Face Uphill Battle

For decentralized compute networks, the challenge is not just technical but also economic. Most DePIN projects operate on token-based incentives, rewarding contributors with native tokens in exchange for providing compute power. This model works well for idle capacity but struggles to attract the kind of institutional capital that Nvidia is now courting. Token volatility can make it difficult for projects to offer fixed pricing, and the lack of service-level agreements is a dealbreaker for many enterprise customers.

There are also regulatory hurdles. Decentralized networks that transfer computing power across borders may face compliance issues related to data sovereignty and export controls. Nvidia’s AI factories, by contrast, can be built in specific jurisdictions with clear legal frameworks and data residency guarantees. This is particularly important for industries like healthcare and finance, where strict regulations govern data handling.

Despite these obstacles, decentralized compute networks are not without hope. Some projects are exploring hybrid models that combine centralized and decentralized elements, allowing users to choose between speed and censorship resistance. Others are focusing on niche use cases where privacy is paramount, such as confidential AI inference and encrypted computation. These areas may not require the massive scale of frontier AI factories, giving DePIN networks a defensible niche.

The Future of Compute is Being Decided Now

Nvidia’s latest initiative is a clear bet that the future of compute will be centralized, capital-intensive, and heavily financed by traditional financial institutions. By making AI compute a bankable asset, the company is creating a feedback loop: more investment leads to more infrastructure, which attracts more customers, which generates more revenue, which attracts even more investment. This virtuous cycle will be difficult for decentralized competitors to break into.

At the same time, the growing demand for AI is not a zero-sum game. Even as Nvidia and its Wall Street partners build massive AI factories, there will still be a need for edge computing, privacy-preserving protocols, and community-operated data centers. The question is whether decentralized networks can scale sufficiently to remain relevant and capture a meaningful share of the market.

The next few years will be critical. If Nvidia successfully mobilizes $500 billion in financing, the centralized AI infrastructure buildout will accelerate at a pace that decentralized networks simply cannot match. But the inherent differences in trust models, governance, and accessibility may still preserve a role for crypto-powered compute, particularly among users who prioritize openness and resilience over raw performance. For now, Nvidia’s push appears set to accelerate the consolidation of compute power in centralized hands, leaving decentralized networks to carve out a smaller but still meaningful niche.

Source: Coindesk News

Phone App

The Phone app in iOS 27 receives one major update, but it is a substantial one. Called Call Context, this feature is designed to bridge the gap between your incoming and outgoing phone calls and the information buried in your Mail app. When you dial a business or an organization, Call Context automatically scans your inbox for relevant details and presents them directly on the call screen. Imagine calling an airline to check your flight status: with Call Context, your confirmation code and flight details will appear as soon as the call connects. Similarly, if you're calling a hotel to confirm a reservation, the app can show your reservation code and check-in date without requiring you to search through your email while on the phone.

How Call Context Works

The underlying technology is deceptively simple yet powerful. Call Context does not listen to your conversation or analyze your speech. Instead, it works by identifying the phone number you are calling and cross-referencing it with the content of your Mail app. If the system finds emails from that company containing booking confirmations, itineraries, or reservation codes, it will surface that information in a clean, unobtrusive panel that appears during the call. This is a classic example of Apple Intelligence, the company's on-device machine learning framework, being used to save time and reduce friction.

Privacy remains a cornerstone of this feature. Call Context processes everything entirely on-device, meaning your email content, call history, and the extracted data never leave your iPhone. This local processing not only protects your personal information but also ensures that the feature works offline. It is worth noting that Call Context requires an iPhone 15 Pro or newer model, as the on-device machine learning models are computationally intensive and need the Neural Engine power found in those devices. The feature is also available when making calls from your iPad or Mac, as long as the devices are signed into the same Apple account and meet the hardware requirements.

FaceTime

FaceTime, Apple's video and audio calling service, also sees a few notable refinements in iOS 27. The headline feature is called Dual Capture, and it finally brings a multi-camera video calling experience to the iPhone. This feature is exclusive to the newest hardware, specifically the iPhone 17, iPhone 17 Pro, iPhone 17 Pro Max, and iPhone Air. With Dual Capture, you can activate both the front-facing camera and the rear camera simultaneously during a FaceTime call. The person on the other end will see a split-screen or picture-in-picture view that includes your face and whatever the back camera is pointed at.

Using Dual Capture

Enabling Dual Capture is straightforward. During a FaceTime call on a supported iPhone, simply tap the camera flip button. This will switch the main video feed to the rear camera, and a small inset window will appear showing your front camera feed. You can tap the inset to swap the main feed, giving you control over which camera is prominent. If you want to show only the back camera feed, you can tap the front camera view to turn it off entirely. This flexibility is particularly useful for showing someone a live view of your surroundings while maintaining face-to-face contact, such as when giving a virtual tour of a new apartment, showing a product from multiple angles, or letting a friend virtually experience a live event.

One of the most impressive aspects of Dual Capture is its compatibility on the receiving end. While only iPhone 17, iPhone 17 Pro, iPhone 17 Pro Max, and iPhone Air can send a dual video stream, any device running iOS 27 can receive and display the dual feed. Since decoding video is a less demanding task than encoding and streaming two video streams simultaneously, there are no hardware requirements for the receiving device. This means you can call a friend with an older iPhone, such as an iPhone 13, and they will still be able to see both your face and your rear camera view seamlessly.

Additional FaceTime Improvements

Beyond Dual Capture, iOS 27 introduces several under-the-hood improvements that enhance the overall FaceTime experience. Connection Quality is a broad term used by Apple to describe a series of network optimizations that reduce stuttering and pixelation when your connection is less than ideal. FaceTime now uses adaptive bitrate streaming more aggressively, meaning it can dynamically lower the video resolution to keep the call audio clear and the video smooth, even on slow networks. This is especially beneficial for users who frequently take calls in areas with poor cellular coverage.

Wi-Fi to Cellular handoff has also been improved. In previous versions of iOS, transitioning from a Wi-Fi network to a cellular network could cause a brief drop or a noticeable pause in the call. iOS 27 uses predictive network monitoring to anticipate when your signal is weakening and prepares the transition in advance. The result is fewer dropped calls and less noticeable interruptions when you walk out of your house, leave an office building, or move between access points. These improvements are subtle but contribute to a much more reliable calling experience.

Two accessibility and safety features round out the FaceTime update. Live Captions, which automatically transcribes spoken dialogue in real time, now supports Traditional Chinese. This is a significant addition for native speakers of Cantonese or Mandarin who rely on captions for accessibility or for those who prefer to read along during calls. Communication Safety, first introduced in iOS 15 as a way to blur nude images sent to children, has been expanded in iOS 27. It now has the ability to detect and block graphic content such as gore and violence in both FaceTime and Messages. This is part of Apple's continued effort to make its platforms safer for younger users, and it extends the existing protection beyond explicit imagery.

iOS 27 Availability

iOS 27 is currently in beta testing, with developers and public beta testers able to download the latest builds through the Apple Developer Program or the public beta channel. The beta has gone through several iterations, and Apple is likely to release a few more beta versions before the final public release. The software is expected to launch this fall, coinciding with the release of the new iPhone models. As is typical for Apple, the upcoming iPhone lineup will ship with iOS 27 pre-installed, and older iPhones that are compatible with the update will receive it via over-the-air update.

For users with older iPhones, it's important to note that while some features like Dual Capture are hardware-exclusive, many of the iOS 27 improvements will be available on a wide range of devices. Call Context requires an iPhone 15 Pro or later, but other features such as improved connection quality and expanded Communication Safety are expected to be available on all models that support iOS 27. This tiered approach to feature availability is consistent with Apple's strategy in recent years, where the most advanced machine learning features are reserved for the newest chips, while general performance and security updates are broadly distributed.

The FaceTime and Phone app changes in iOS 27 may not be the flashiest additions in this year's update, but they represent a clear direction Apple is taking: using on-device intelligence to make communication more efficient and reliable. Call Context turns your phone into a smart assistant that knows why you're calling before you even speak, while Dual Capture opens up new possibilities for creative and practical video calls. Combined with the network reliability improvements and safety enhancements, these updates ensure that the two most essential apps on your iPhone are better than ever when iOS 27 arrives this fall.

Source: MacRumors News

How It Started

Long before ChatGPT became a household name, educators and editors frequently used anti-plagiarism tools to verify the originality of written work. These systems compare a manuscript against vast databases of web pages, academic papers, and published books to identify matching sentences and phrases. Tools like Turnitin assign a similarity percentage, indicating how much of a student's writing overlaps with existing sources. But even these traditional plagiarism detectors have faced criticism over false positives and ambiguity about whether duplicated passages were intentional or accidental. Some educators have already abandoned them for these reasons.

Now, the hunt for copied content has evolved into a war on AI-generated text. As quickly as students adopted ChatGPT, Google Gemini, and Microsoft Copilot, teachers adopted so-called AI detectors with equal speed. A survey from the Center for Democracy and Technology found that 43 percent of sixth to 12th grade teachers in the US regularly used AI detectors between 2024 and 2025. Some universities that already used Turnitin discovered that the service automatically enabled its AI detection feature when the tool launched in 2023, without explicit consent from instructors or students.

Unlike traditional plagiarism checkers, AI detectors like GPTZero, Pangram, and Turnitin's proprietary system do not match text against a database. Instead, they rely on their own AI models to estimate the likelihood that a piece of writing was generated by a machine. This process is arguably even murkier than matching text on the web. GPTZero explains that its algorithm analyzes a text's wording, rhythm, and structure, and looks for patterns in length and tone that appear more frequently in AI-written content. This relatively subjective evaluation is not as solid as a direct text match, and it can easily be confused by writers who use English as a second language. Despite these limitations, Turnitin claims that its AI detector falsely flags less than 1 percent of human-written content, while Pangram asserts a false positive rate of just 1 in 10,000. GPTZero similarly claims a very low rate of mistaking human content for AI.

How It's Going

People online are already accusing each other of "sounding like AI," but the ready availability of AI detection tools is only adding fuel to the LLM witch hunt. In several high-profile cases, AI writing accusations have directly impacted people's livelihoods and reputations. Last month, the publisher Minotaur dropped a $2 million book deal over concerns that its author, Jerry Falade, used AI — something he vehemently denies. The accusation alone was enough to derail his career, even though no definitive proof was presented.

There is also the case of Thierry Rignol, a French national who sued Yale University after a professor accused him of using AI to write portions of his final exam. The accusation resulted in a failing grade and a one-year suspension. The professor had used GPTZero to scan Rignol's writing for signs of AI. The lawsuit argues that AI surveillance and detection tools are known to unfairly target non-native English speakers like Rignol. In February, a student at Adelphi University won a lawsuit against the school after a professor similarly claimed he used AI to write an essay. The lawsuit did not specify which AI tool was used, but Adelphi University has a licensing agreement with Turnitin.

A 2023 Stanford study found that AI detectors falsely flagged essays written by non-native English speakers as AI more often than essays by native speakers. Many services still claim their tools are accurate when dealing with text written by non-native speakers, but the study contradicts this. These tools may also be biased against neurodivergent writers, whose unique writing patterns can be misinterpreted as machine-generated.

As the University of California, Los Angeles points out, AI detection tools are trained to pick up on patterns that could indicate AI use, such as repetitive terms and phrases, text that sounds too formal or informal, and nonsensical phrasing. Some tools, like QuillBot, also measure the "unpredictability" of text. According to UCLA, AI tends to make the most "obvious" or most common language choices compared with human-produced writing. Detectors may also flag consistent sentence structure as a sign of AI. But these measurements are not indicative of AI on their own, as some people may simply have a writing style with these qualities.

Even Turnitin, which touts low false positive rates, maintains that its tool "may not always be accurate" and should not be used to take action against a student. Grammarly warns that users "should never rely on the results of an AI detector alone," while GPTZero says "no AI detector can ever truly be 100% perfect." OpenAI even shut down its own AI writing detector in 2023 due to low accuracy. Despite these disclaimers, AI writing accusations are still being flung across the web.

Last week, in a video broadcast to the more than 3.5 million followers across his social channels, Jack Osbourne accused journalist Kat Tenbarge of using AI to write an article for Rolling Stone. He flaunted the results from an AI detector called Getsolved as "proof" of his claim. Tenbarge has refuted the claim in a video and a post on her website, but Osbourne has not retracted his accusation or deleted the video, leaving Tenbarge to deal with online trolls. This is just one of many examples of false accusations, from writers to students, with accusers often ignoring the disclaimers that come with these tools.

What Happens Next

The uncertainty surrounding AI detectors is enough for some educational institutions to stop using them altogether. Yale University, Johns Hopkins University, Vanderbilt University, Georgetown University, and others have disabled or restricted the use of AI detection tools. The Massachusetts Institute of Technology also warns that "AI detectors don't work."

Instead of relying on tools to weed out AI, many schools are encouraging educators to rethink their lessons. The University of Chicago suggests telling students to slow down their reading, breaking up longer writing assignments, and requiring students to reflect on their work. Stanford University says professors can consider holding assessments in classrooms, while MIT advises professors to leave room for students to disclose whether they used AI for help on an assignment, without penalty.

With AI becoming more prevalent in and outside the classroom, efforts to suss out what is written by a human or a machine are ramping up as well. Some online platforms are exacerbating suspicions surrounding AI use. Substack has built Pangram into its app, allowing users to scan blogs for suspected AI-generated content, while LinkedIn added a "seems like AI slop" button on posts. The result is a new era of distrust, where readers constantly question whether what they are reading is AI and real human writers try their best not to sound like it.

By the Way

• The Authors Guild is helping writers get ahead of accusations by giving them "Human Authored" certifications. There are also Not by AI and Written by Human badges users can add to their work online.
• Wikipedia created a guide to help editors spot AI writing, which includes looking for writing that "puffs up" the importance of a topic or provides "superficial analysis of information." The site has also banned AI-generated articles.
• The New York Times has a quiz that asks readers to look at five pairs of passages and choose which blurb they like better — the catch is that one of them is written with AI.
• Inside Higher Ed spoke to some educators about how they are approaching AI in school, with one lecturer raising concerns about the costs and resources that could go into AI-proofing assignments.
• The fanfiction community is having its own internal struggle over how to determine which stories are generated by AI, as detailed by a report from a tech news outlet.

The growing reliance on AI detectors is not just a technical issue; it is a social one. As these tools become more embedded in education, publishing, and social media, the risk of false accusations rises. The disclaimers from developers are often buried in fine print, while the confidence of accusers is loud and public. Schools are beginning to adapt by emphasizing process-oriented assignments and classroom discussions, but the broader culture is still caught in a cycle of suspicion. Until the technology improves or trust is restored through better practices, the burden of proof will continue to fall on individuals who must prove their humanity.

Source: The Verge News

Google’s Pixel 11 series is finally official, and while the hardware changes may seem subtle at first glance, the software story is far more ambitious. The new lineup includes the Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL, and Pixel 11 Pro Fold, all shipping on August 20th. With a heavier focus on camera intelligence, personalization, and creator-friendly tools, Google is positioning the Pixel 11 family as more than just a spec bump.

Pixel 11: refined design, higher price

The standard Pixel 11 starts at $899 for a 256GB model, which is a $100 increase over last year’s base configuration. Google has eliminated the 128GB storage tier across the lineup, and all phones now come with at least 12GB of RAM. The price hike is largely attributed to broader industry cost pressures, but Google is hoping the refinements justify the premium.

Visually, the Pixel 11 keeps the signature camera bar, but it’s now made from a single piece of glass and is over 40 percent thinner. When paired with Google’s official cases, the camera bar sits completely flush with the back panel. The phone is available in purple, green, black, and a bright magenta option that stands out in person.

Under the hood, the Tensor G6 chipset powers the new devices, offering up to 20 percent better efficiency. This translates to longer battery life, with Google quoting up to 30-plus hours on a single charge across the lineup. The camera system on the Pixel 11 also gets a notable upgrade. The 48MP main sensor now uses a larger 1/1.56-inch sensor, which Google says captures 56 percent more light. The 5x telephoto lens can now digitally zoom up to 30x, thanks to the improved image signal processor inside the Tensor G6.

Camera Looks: a new way to capture photos

One of the most intriguing additions to the Pixel 11 is the Camera Looks feature. Similar to film recipes found on Fujifilm cameras, Camera Looks gives users stylized presets that can be customized. Unlike simple filters, however, these looks change how the phone processes the photo at the sensor level. Depending on the selected look, the Pixel may reduce frame stacking and sharpening, resulting in a more natural or artistic image straight out of camera.

During early briefings, the Natural mode impressed by lowering sharpness and contrast, making photos look far more true to life. This is a significant departure from the overly processed look that smartphone cameras often produce. Google’s camera lead explained that the goal is to give users more creative control over the final image without requiring manual editing.

Magic Capture and Creator Suite

The camera app also gains a new Magic Capture mode, which blends features of photo and video recording. The phone automatically picks out short clips and stills from a single continuous recording, aiming to capture fleeting moments without user intervention. In initial testing, the feature was still rough, producing limited results from a short recording, but Google believes it will excel at events like children’s birthday parties where you might want a mix of photos and video.

For those focused on social media, the new Creator Suite is built directly into the camera. It includes social media framing guides, speech enhancement tools, a teleprompter, and automatic video folder organization. This integration signals Google’s intent to compete for small creators who might otherwise turn to dedicated camera gear.

Qi2.2 wireless charging and other improvements

All Pixel 11 phones support Qi2.2 wireless charging, an upgrade from Qi2 last year. The new standard promises 25 percent faster wireless charging speeds, reaching up to 25W on the Pixel 11, compared to 15W previously. Last year’s Pixel 10 had mixed results with third-party chargers, sometimes dropping to just 5W, so this year’s improvement should address some of those complaints.

Pixel 11 Pro and Pro XL: brighter displays and better zoom

The Pixel 11 Pro and Pro XL sit higher in the lineup, starting at $1,099 with 256GB of storage and 12GB of RAM. That’s a reduction from 16GB of base RAM last year, though you can still get 16GB if you choose the 512GB or 1TB models. The Pro phones also see a price increase of $100.

The Pro displays have been upgraded to a peak brightness of 3,600 nits, which Google says is the brightest on any smartphone. The main 50MP sensor is new and more power-efficient, and the 48MP telephoto lens now uses a larger 1/1.95-inch sensor for a 30 percent boost in light sensitivity. The camera’s zoom range has also been extended to 120x digital zoom, powered by the Tensor G6 ISP.

Night Sight has also received a major speed boost. Google says low-light capture is up to 4.5x faster, with average improvements around 3x. This makes it easier to take handheld night shots without asking subjects to hold still for long periods. You can still use longer exposures up to four minutes if you want maximum light.

HiLight LED: a new way to stay connected

The flashiest hardware addition to the Pro models is the HiLight ambient LED. This RGB LED ring replaces the temperature sensor on the camera bar. When you’re talking to Gemini, the LED glows blue. You can also assign it to favorite contacts, causing the ring to light up when they call through the default Phone app or WhatsApp.

While the feature is visually appealing, its usefulness remains limited. Google says more functionality will arrive later, but for now, it’s more of a novelty than a must-have. Still, it’s a unique way to get notifications at a glance without looking at the screen.

The Pixel 11 Pro and Pro XL come in olive, light blue, a striking “canyon” orange, and black. Most models have glossy side rails, but the black option features matte rails for a more subdued look.

Pixel 11 Pro Fold: durability and refinement

Google’s latest foldable, the Pixel 11 Pro Fold, is positioned as the most durable foldable the company has made. It uses a new composite back material that Google claims is nearly impossible to crack. Combined with a thicker ultra-thin glass display, the phone is rated 3x more durable than last year’s model.

The inner bezels have been reduced significantly, giving it the thinnest bezels ever on a Pixel Fold. Peak brightness matches the Pro models at 3,600 nits. The Fold inherits the 48MP main sensor from the Pixel 11, offering a 56 percent improvement in light sensitivity, and supports Qi2.2 for 20 percent faster wireless charging.

The Pixel 11 Pro Fold starts at $1,899 with 16GB of RAM and 256GB of storage. That’s also a $100 increase over last year. It comes in olive and black.

Software and ecosystem integration

Beyond the camera, the Pixel 11 series runs Android with deeper Gemini integration. The phones are designed to leverage on-device AI for various tasks, including more natural conversations with the assistant and context-aware suggestions. Google is clearly betting on software to differentiate its devices in a crowded market.

The Pixel 11 also continues Google’s focus on computational photography, using the Tensor G6’s ISP to enable new features like the extended zoom ranges and faster Night Sight. The new Sensor Look feature set is a clear departure from the one-size-fits-all approach of previous Pixels, giving users more control over how their photos are processed.

Availability and pricing

All four Pixel 11 models are available for preorder starting today, with shipments and in-store availability beginning August 20th. The entry-level Pixel 11 is priced at $899 for 256GB, and storage can be upgraded at additional cost. The Pro models start at $1,099, while the Fold starts at $1,899. Preorder promotions include gift cards worth up to $350 at select retailers.

Google is also offering a range of first-party cases, including a new design that makes the camera bar completely flush on the standard Pixel 11. With three different camera bar thicknesses across the lineup, users can choose the look and feel that suits their preferences.

The Pixel 11 series represents a mature iteration of Google’s smartphone vision. While the hardware changes are incremental, the software innovations and camera processing advancements could make a meaningful difference in daily use. Whether that justifies the higher price remains to be seen, but the Pixel 11 lineup is arguably the most feature-packed Pixel family to date.

Source: The Verge News

Pixel 11 lands with new colors, faster performance, and a higher starting price

Google refreshed its entry-level flagship with the Pixel 11, a phone that keeps the same overall design language as the Pixel 10 but introduces a splash of new color options. The new palette includes a lilac-like frost, hibiscus pink, pistachio green, and obsidian black. While the hardware is familiar, the internals have received a significant upgrade. The Pixel 11 runs on Google’s new Tensor G6 chip, which promises 25 percent faster web browsing and 15 percent faster app loading compared to its predecessor. Google also claims the battery now lasts over 30 hours, and wireless charging is 25 percent faster than before.

The camera system has also been refreshed. The Pixel 11 features a camera bump that is more than 40 percent thinner than the one on the Pixel 10, which should make the phone feel more comfortable in the hand. The rear setup includes a 48MP wide lens, a 13MP ultrawide lens, and a 10.8MP 5x telephoto lens, while the front selfie camera comes in at 10.5MP. Google says the 48MP wide lens has 56 percent more light sensitivity for better low-light shots, and the 5x telephoto lens supports up to 30x Super Zoom.

The most notable change for many consumers will be the price. The Pixel 11 now starts at $899, a $100 increase over the Pixel 10’s launch price. However, the base configuration now includes 256GB of storage and 12GB of RAM, rather than the 128GB of storage found in the Pixel 10’s entry model. The Pixel 11 will be available starting August 20th.

Pixel 11 Pro and Pro XL introduce Google’s new HiLight LED

Moving up the lineup, the Pixel 11 Pro and Pixel 11 Pro XL come with a novel hardware addition called HiLight. This is a small LED indicator built into the camera bar that lights up for specific events, such as Gemini interactions or incoming calls from favorite contacts. Google says the LED will gain more capabilities over time, including notifying users of messages from frequent contacts. It’s a subtle but potentially useful feature for those who keep their phones face down or on a desk.

The Pro models also receive a major display upgrade, with a maximum brightness of 3,600 nits. Inside, they share the same Tensor G6 chip as the standard Pixel 11. The camera system is more advanced, featuring a new 48MP telephoto lens with up to 120x Pro Zoom, a new 50MP wide lens, a 48MP ultrawide lens, and a 42MP selfie camera. Google says the Pro devices offer over 30 hours of battery life and more than 29 percent faster wireless charging compared to the Pixel 10 Pro and Pro XL.

Pricing for the Pixel 11 Pro starts at $1,099 with 256GB of storage and 12GB of RAM. The Pixel 11 Pro XL starts at $1,299 with the same specs. Those who need more space can opt for 512GB or 1TB models, which come equipped with 16GB of RAM. The Pro and Pro XL are available in four colors: canyon, fog, olive, and matte obsidian.

Pixel 11 Pro Fold launches with durability upgrades

Google’s foldable flagship, the Pixel 11 Pro Fold, also got a refresh with attention focused on durability and display quality. The company claims it now has the thinnest bezels ever on a Fold and is three times as durable as the Pixel 10 Pro Fold. It carries an IP68 water and dust resistance rating and uses a back material that Google says is nearly impossible to crack. That’s a notable improvement for a foldable, which often has more fragile moving parts.

The displays are 20 percent brighter than last year’s model, with the inner display reaching up to 3,600 nits, matching the Pro phones. Under the hood, the Fold runs on the Tensor G6 chip and offers up to 24 hours of battery life. The camera system includes a new 48MP wide lens, plus a 10.5MP ultrawide lens and a 10.8MP telephoto lens with up to 30x Super Zoom.

The Pixel 11 Pro Fold costs $100 more than last year’s model, starting at $1,899. The base configuration includes 256GB of storage and 16GB of RAM. It comes in two colors: olive and black.

Pixel phones are getting some new camera and creator features

Beyond the hardware upgrades, Google is adding a range of new software features designed to make the Pixel lineup more capable for photography and video creation. The Camera app gains a new capture mode called Magic Capture, which can pull still photos and clips directly out of video recordings. This is useful for quickly grabbing a perfect frame from a moving scene without having to take a separate photo.

There’s also a new feature called Camera Looks, which offers customizable, stylized photo presets that go beyond simple filters. These Looks change how the phone processes the image, with styles such as Minimal, Editorial, and “Digi,” which mimics the look of a digital camera. The Pixel 11 Pro also introduces Pro Stable Video, a feature that Google says can record at up to 8K resolution with improved stabilization.

For content creators, the Camera app now includes a dedicated Creator tab with a “Creator Suite.” This suite offers tools for trimming and reordering videos, an onscreen teleprompter, and improved Speech Enhancement. Together, these features make the Pixel line a more viable option for vloggers and social media creators who want to produce polished content without needing extra equipment.

Pixel Watch 5 gets a suite of new health features

The Pixel Watch 5 looks almost identical to its predecessor, but under the surface it brings a host of new health and fitness capabilities. One of the standout additions is Breathing Emergency Detection, a follow-up to last year’s Loss of Pulse feature. This new tool can trigger an emergency call if the watch detects a significant, sustained drop in blood oxygen saturation (SpO2). It’s designed to offer peace of mind for users who may experience respiratory issues during sleep or at other times.

In addition, the Pixel Watch 5 can track nightly sleep breathing quality and summarize the data in monthly reports. Starting in September, Google will also introduce Blood Pressure Trends and Insulin Resistance Trends, giving users more insights into metabolic health. On the workout side, the watch offers AI-generated strength training plans and the option to track reps directly from the wrist.

Google claims the Pixel Watch 5 has the most accurate GPS route tracking for exercise, outperforming the Apple Watch Ultra 3 and Garmin Fenix 8 Pro. As with the Pixel phones, there is a price increase: the Pixel Watch 5 now starts at $399, up $50 from the prior generation. It’s available in two sizes — 45mm and 41mm — and four colors: black, white, olive green, and pink. There’s also a Steph Curry special edition that costs $579 and launches later, on September 3rd.

Google’s launching its own version of Apple’s AirTag

Google finally entered the item tracker market with the Pixel Tag, a direct competitor to Apple’s AirTag. The device launches on November 11th and will cost $29 for a single unit, or $99 for a pack of four. That pricing matches the AirTag exactly, making it an easy alternative for Android users. The Pixel Tag has an oblong shape rather than the AirTag’s circle, measuring 1.1 inches by 1.8 inches (the AirTag is 1.26 inches across). It’s IP67 water and dust resistant, and Google says its replaceable CR2032 battery lasts over a year.

The Pixel Tag is designed to integrate with Google’s Find My Device network, which should make it simple to locate misplaced items like keys, bags, or luggage. With the AirTag already having a strong presence in the tracking category, the Pixel Tag gives Android users a compelling option that leverages Google’s ecosystem.

Pixel Buds Pro 2 are getting improved ANC and a new color

The two-year-old Pixel Buds Pro 2 are also getting some attention. Google is releasing the earbuds in a new olive green colorway on August 12th, which matches the olive option available for the Pixel 11 Pro and Pro Fold. In September, the Buds Pro 2 will receive a firmware update that aims to improve active noise cancellation (ANC). Since the Buds Pro 2 have already been well-received for their sound quality and comfort, these updates should help extend their lifespan and keep them competitive with newer rivals.

In addition to the new color and ANC improvements, the Buds Pro 2 remain a solid choice for Android users, offering useful features like gesture controls, seamless pairing with Pixel devices, and multi-device connectivity. Google’s decision to update the earbuds with new software features shows that the company is committed to supporting its hardware beyond the initial release.

Together, the announcements from the Made by Google event paint a picture of a company that is focused on integrating hardware and software tighter than ever. With the Pixel 11 series, Google is pushing performance and camera capabilities, while the Pixel Watch 5 aims to be a proactive health companion. The Pixel Tag enters a growing market, and the Pixel Buds Pro 2 receive a timely refresh. For anyone in the market for new Android devices, this year’s Pixel lineup offers a wide range of options to fit different needs and budgets.

Source: The Verge News

The announcement came during Google’s Made by Google 2026 event, alongside the new Pixel 11 smartphones and Pixel Watch 5. While those devices are already available for preorder and ship next week, the Pixel Tag will not be available until November 11th. That staggered launch gives Google time to prepare the accessory ecosystem and ensures that early Pixel 11 owners have a new first-party tracker to pair with their phones.

Design and hardware

The Pixel Tag is a slim oblong device, noticeably longer but thinner than Apple’s AirTags. It comes in a single grayish color that Google calls Fog. The device weighs approximately 12 grams, or 0.4 ounces, making it light enough to attach to keys, backpacks, luggage, or other everyday items. As with the AirTag, the Pixel Tag has no built-in hook or clip, so users will need to buy third-party accessories to attach it to objects.

The Tag includes a single physical button that triggers a sound on your phone, which is useful for the rare situation where you have the Tag but have lost your primary device. This is a slightly different approach from the AirTag, which does not have a button at all. The Pixel Tag also has an IP67 rating, meaning it is protected against dust and temporary immersion in water, so it should be safe in rain or with minor splashes.

Power is provided by a standard CR2032 coin battery, which Google says should last for over a year. The replaceable battery is a welcome feature for users who plan to use the Tag for years. It also matches the industry standard set by the AirTag, which uses the same battery type.

Precision tracking with UWB and Bluetooth Channel Sounding

The Pixel Tag is one of the most capable trackers on the market because it supports two distinct types of precision tracking. First, ultra wideband (UWB) allows the Find Hub app to display both distance and direction to the tracker. This is similar to the Precision Finding feature found in Apple’s AirTag. UWB works by measuring the time it takes for radio waves to travel between the phone and the Tag, enabling centimeter-level accuracy. However, UWB requires a phone with a dedicated UWB chip.

Second, the Pixel Tag supports Bluetooth 6.0 Channel Sounding, a newer development that can display the exact distance to the tracker but not its direction. Channel Sounding uses a phase-based technique to measure distance over Bluetooth, making it more accurate than traditional Bluetooth RSSI-based tracking. Unlike UWB, Channel Sounding works with any phone or tablet that includes Bluetooth 6.0, even if it lacks a UWB chip. This broadens the range of devices that can take advantage of precise distance tracking.

The combination of UWB and Channel Sounding makes the Pixel Tag more versatile than many existing trackers. While UWB offers the full directional experience on high-end phones, Channel Sounding provides a fallback for mid-range devices. The Moto Tag 2 is currently the only other tracker that supports Channel Sounding, giving Google and Motorola an early lead in this area.

Google Find Hub integration

The Pixel Tag is designed to work with Google’s Find Hub network, which leverages the Bluetooth capabilities of Android devices worldwide to help locate lost objects. Find Hub has been available for more than two years, and Google has steadily improved it since a rocky launch. In an independent test conducted last year, the network was still less precise than Apple’s Find My network, especially in crowded urban areas. However, Find Hub has grown significantly as more Android devices have joined the network, and the Pixel Tag is expected to benefit from that scale.

When a Pixel Tag is lost, nearby Android devices running Google Play services can anonymously report its Bluetooth signal back to the owner. This crowd-sourced approach is the same fundamental idea as Apple’s Find My network, but Google has added privacy safeguards to prevent unwanted tracking. The Find Hub app will show you the last known location and, when you are close, use UWB or Channel Sounding to guide you to the exact spot.

Device compatibility and ecosystem

Google device owners will need a Pixel 10 or later to take advantage of Channel Sounding, although the Pixel 11 series also supports it. For UWB, users will need a Pixel Pro model since the Pixel 6 Pro. Notably, the base model in the Pixel 11 series still lacks UWB, so those users will have to rely on Channel Sounding for precise distance measurements. This means that the Pixel Tag is most useful for people who own a Pixel Pro smartphone or a high-end device from another manufacturer that supports UWB.

The Pixel Tag can also be tracked using a Pixel Watch. Any Pixel Watch model since the 4 will support Channel Sounding, while the Pixel Watch 2 and later include UWB. This makes the Tag a versatile accessory for the entire Google hardware ecosystem. In addition, Google has integrated Gemini, its AI assistant, so you can ask Gemini to ring the Tag’s speaker for you. This voice control feature adds another layer of convenience, especially when your phone is not in your hand.

Competitive landscape

With the Pixel Tag, Google is directly challenging Apple’s AirTag 2, which also retails for $29 or $99 for a four-pack. The AirTag 2 has been the market leader in the tracker segment since its release, thanks to Apple’s massive Find My network and its deep integration with the iPhone. Google’s answer is a tracker with similar pricing and arguably more advanced positioning technology, given the inclusion of Channel Sounding alongside UWB.

The Pixel Tag also faces competition from Samsung’s Galaxy SmartTag series and various third-party trackers like the Moto Tag. However, Google’s decision to standardize on Bluetooth 6.0 Channel Sounding could accelerate the adoption of this new technology across the Android ecosystem. If more device makers implement Bluetooth 6.0 in their phones, the Pixel Tag will become a more compelling choice for users who do not own a Pixel phone.

The timing of the Pixel Tag is notable. Google has spent years building out its Find Hub network and now finally has the hardware to showcase it. The Tag is a strategic piece of the Pixel ecosystem, designed to keep users within Google’s services and encourage more people to buy Pixel devices. While the tracker itself is simple, it represents a mature move from Google in a category where it has lagged behind Apple.

The Pixel Tag ships on November 11th, with preorders expected to open closer to the launch date. Google will also sell a four-pack for $99, making it easy for families or travelers to track multiple bags and devices. With its dual-mode positioning and affordable price, the Pixel Tag is poised to become one of the most versatile item trackers on the market, though its success will depend on the continued growth of the Find Hub network and the adoption of Bluetooth 6.0 across Android devices.

Source: The Verge News

The ideas are not entirely new. Zuckerberg has touched on similar themes in earnings calls and long-form essays for years. But this latest post is probably the most detailed public explanation of how he thinks AI will change society. It has been described in some corners as an anti-doomer document, but that label misses the point. The essay is less a response to catastrophic risk and more an explanation of why Zuckerberg is personally excited about the technology.

That excitement, however, keeps colliding with reality. Even as he promises a future of abundance, he offers reminders of why the public remains anxious about AI. For anyone who has followed the last decade of social media, the pattern is familiar: a tech visionary describing a technology in ideal terms, while ignoring how the same tools can be abused, misused, or redistributed in ways that are not purely positive.

A Trust Deficit That AI Hasn't Reset

Social media remains a sensitive subject across the technology industry, and Zuckerberg and his company carry much of that baggage. Polling has shown that a large majority of Americans believe social media has been harmful to democracy. A recent survey cited in the original piece put that number at 64 percent, with a similar share calling for stronger regulation. That concern spans partisan lines, making it one of the few issues on which Americans seem to agree.

The legal and regulatory pressure has also intensified. Just over the weekend, a court fined the company more than half a billion dollars in a case related to harm to children. The details of that case are complicated, but the underlying message is not: the public has seen what happens when platform growth is prioritized over user safety, and they do not want a repeat performance with AI.

This history matters because it is one of the main reasons people are nervous about the social impact of AI. It is not necessarily that people think AI is inherently evil. It is that they no longer trust the people building it to handle the consequences. When Zuckerberg writes about superintelligence with the same abstract vocabulary he once used for free speech, he is not reassuring anyone. He is reminding them of exactly what made them skeptical in the first place.

Hazy Ideals and Hard Trade-offs

Much of the manifesto is written in philosophical language. Zuckerberg argues that as everyone gains more powerful tools, each person will become more capable of shaping the future. He suggests that people and institutions with competing interests naturally check and balance each other, leading to positive outcomes. That is a comforting story, but it is not an obvious one. There are countless examples of conflicting interests producing gridlock, exploitation, or outright harm. Treating balance as an automatic outcome is less a conclusion than a hope.

The strange thing is that these arguments could have been made in a more constrained way. Zuckerberg is not describing an abstract future; he is describing a product category that Meta is actively bringing to market. If listeners already distrust the speaker, a manifesto that keeps insisting nothing can go wrong will only deepen that distrust. The lack of humility is not a minor rhetorical problem. It is a signal that the person in charge might not be asking the hard questions.

When the essay turns to specific examples, the gap between vision and reality becomes even clearer.

The Education Example Misses the Obvious Problem

Zuckerberg promises that everyone will have a personalized tutor and coach with a PhD in every subject and unlimited patience. In his telling, students will receive extra help in areas where they need it, and adults will have superintelligent learning assistants that teach new job skills, languages, and hobbies.

But this product already exists in a rough form. ChatGPT, Claude, and Gemini are widely used as consumer chatbots, and people do use them to learn. But the most common educational use of these tools is not tutoring. It is cheating. Students use AI to write essays, solve problem sets, and complete assignments they are supposed to do themselves. Because there is no reliable watermarking system, teachers often cannot prove that AI generated a given text.

That is a real harm happening right now, not a hypothetical one. It is a textbook case of unintended consequences. A technology designed to make information more accessible ends up making assessment less meaningful. The fact that one of the most powerful people in the world does not acknowledge this problem is exactly why people worry. When a leader talks only about positive use cases and ignores the messy reality, it becomes difficult to believe they are prepared for problems at larger scale.

The Legal System Example Avoids Complexity

Zuckerberg also offers a thought experiment about superintelligent lawyers. He argues that if only one person had such a lawyer, that person would have an unfair advantage in court. But if everyone has a superintelligent lawyer, justice would be carried out more fairly and efficiently, because the imbalance in skills and resources would disappear.

There is a kernel of truth to that vision. Access to legal help is deeply unequal, and AI could lower the barrier for people who cannot afford traditional representation. But more access does not automatically mean better outcomes. It could also mean a flood of automated filings and vexatious lawsuits, overwhelming courts that are already overburdened. It could add complexity to an administrative system that is already a bureaucratic hellscape. And a superintelligent lawyer that is wrong could be just as dangerous as one that is right, especially if litigants do not understand how the system works.

The thought experiment assumes a level playing field that does not exist anywhere else in society. Every step of the legal system has gatekeepers, incentives, and structural flaws that are not erased simply by giving everyone a tool. Zuckerberg's confidence that everyone having superintelligence will lead to fair outcomes requires a great deal of faith in markets, institutions, and human behavior. That faith has not been earned by the recent history of technology.

Freemium Models and Dynamic Auctions

Perhaps the most revealing section of the manifesto is the description of Meta's existing business model. Zuckerberg promises that everyone will have free or affordable access to AI tools, and that the company will offer free versions available to billions of people. For those who want more compute, he describes a dynamic auction mechanism that will guarantee the lowest possible price while ensuring capacity is used for what people collectively find most valuable.

In broad strokes, the argument about access is defensible. A freemium model does provide more access than requiring everyone to pay upfront. Dynamic markets for spot compute already exist, so the concept is not outlandish. But there is a reason that consumer AI products generally shield end users from the spot price of the compute they use. Surge pricing is a terrible user experience, especially when applied to a tool someone relies on for work or education. If the price of an AI assistant changes hour by hour based on demand, users cannot plan around it. They will feel punished for using the product at the wrong time.

Zuckerberg may know this, and he may have no intention of exposing consumers to dynamic auctions. But the manifesto does not make that clear. Instead, it presents a market mechanism as a fairness feature, even though many people will hear it as another way for the company to extract more money from them. The failure to anticipate that reading is itself a kind of tone-deafness.

Why Trust Is the Missing Variable

There are better ways to talk about AI. Some of the most visible AI executives have learned to acknowledge the dangers of the technology directly, emphasizing their own precautions and arguing that they can be trusted to act responsibly. That approach has its own flaws, and it cannot fully overcome the inherent risks of the technology. But it at least engages with the public's anxiety instead of dismissing it.

Zuckerberg's manifesto goes in the opposite direction. It paints a bright picture and then brushes past the ways things might go wrong. It promises that competition will check and balance, that generous access will solve inequality, and that better tools will make society more just. Those are not technical arguments. They are expressions of faith. And faith is exactly what the public no longer has in the people building these systems.

The AI industry may not always be able to survive its failures. Without trust, every mistake will look like a confirmation of the worst-case scenario. The path forward is not another triumphant essay about superintelligence. It is humility, specificity, and a willingness to sit with the uncomfortable fact that every powerful tool will produce both good and bad results. Until the people leading this industry can say that plainly, they will keep making the case for why people dislike AI.

Source: TechCrunch News

The confirmation comes as the European Union’s AI Act begins to take effect in stages, with its Transparency Code having been in force since August 2. Under this code, AI companies are required to make AI-generated or AI-edited content identifiable by other systems. This means that machine-readable markers must be embedded in the output so that downstream platforms, regulators, and users can detect whether content has been produced by an AI system.

How the Watermarking Works

According to Anthropic’s support page, watermarking will be applied at the model level, ensuring that the marker is present regardless of which Claude product or surface the text originates from. This includes the Claude platform API, Claude, Claude Code, Claude Cowork, and Claude Tag. The company also said it will extend support for older models, though it has not yet specified a definitive timeline for when legacy versions will receive the watermarking capability.

One notable detail is that the watermark is embedded directly into the text itself. Anthropic explains: “Because the watermark is part of the text, it will travel with the text when it’s copied and pasted elsewhere, and may persist through some editing. Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from.” This approach differs from metadata-only tagging, which can be easily stripped when content is copied, re-uploaded, or converted into another format.

The company did not clarify how much editing a user would need to do to remove the watermark. That question remains open, and Anthropic has been asked for additional details. The robustness of any watermarking scheme is critical, as malicious actors may attempt to strip or alter markers. Text-based watermarking is particularly challenging because language can be rewritten while preserving meaning. Unlike image or video watermarks, which can rely on invisible pixel patterns, text watermarking often requires subtle statistical patterns in word choice or syntax that survive basic edits but degrade under significant paraphrasing.

EU AI Act and Transparency Obligations

The EU AI Act is a landmark piece of legislation that introduces a risk-based framework for artificial intelligence systems. Its Transparency Code mandates that providers of general-purpose AI models ensure that their outputs are marked in a machine-readable way. The code also requires AI systems that generate or manipulate images, audio, or video to disclose that the content is artificial or altered. While the full implementation of the AI Act will be phased over several years, the transparency provisions became applicable in early August, prompting companies like Anthropic to act.

Anthropic is not alone in this effort. Other major AI companies, including Black Forest Labs, Google, Meta, Microsoft, OpenAI, and Synthesia, have committed to adhering to the EU’s code. The industry-wide push follows growing backlash from users and civil society groups who have raised concerns about deepfakes, misinformation, and the erosion of trust in digital content. Labeling AI-generated content is seen as a first step toward ensuring accountability and helping people distinguish between human and machine authorship.

The Rise of Content Provenance Standards

The C2PA standard, which stands for Coalition for Content Provenance and Authenticity, is an open technical standard that allows publishers, creators, and platforms to embed information about the origin and history of digital content. C2PA uses cryptographic techniques to sign content metadata, making it tamper-evident. Several tech companies, including Adobe, Microsoft, and Intel, have backed the standard, and it has become a baseline for provenance efforts in the media industry.

However, C2PA metadata can be removed when content is processed or translated. Anthropic’s decision to embed watermarking in the text itself is intended to address this limitation, at least partially. The company’s approach appears to combine model-level watermarking with C2PA for files, creating a dual layer of traceability. This is particularly relevant for text-based AI outputs, which are often copied into emails, documents, and social media posts where metadata may not survive.

Industry Reactions and Parallel Moves

Anthropic’s announcement comes amid a broader wave of watermarking commitments across the AI industry. Last week, AI music platform Suno said it would mark tracks created on its platform after facing a spate of legal challenges from record labels and artists. The music industry has been particularly vocal about the need to distinguish AI-generated audio from human performances, especially in contexts where royalties and copyright are at stake.

In the publishing world, newsletter service Substack recently teamed up with Pangram to flag AI-generated content. Substack’s CEO, Chris Best, used the occasion to call out “Claudefishing,” a term describing people who use AI to generate content and pass it off as their own. This is analogous to “catfishing,” and the term has quickly gained traction in online communities where authenticity is highly valued. Best’s remarks highlight the societal pressure on platforms to take responsibility for AI content detection and disclosure.

The trend is not limited to large corporations. Independent developers and researchers have been exploring watermarking techniques for years, with some proposing methods that embed imperceptible signals in text using language model decoding strategies. These techniques often rely on choosing between statistically equivalent word sequences based on a secret key, which can later be verified. The challenge is to make the watermark robust enough to survive minor edits while undetectable to users who are not specifically looking for it.

What This Means for Users

For everyday users of Claude and other Anthropic products, watermarking will be invisible in most cases. The text will read normally, and there will be no visible marker or banner unless a platform chooses to display the provenance information. The watermark is designed to be detected by automated systems, not by human readers. This could mean that AI-generated content circulated in emails, blog posts, or academic papers may eventually be flagged by plagiarism checkers, content moderation tools, or search engines that adopt the same detection standards.

For businesses and developers using the Claude API, the watermark is automatically included in all outputs for models released after August 2. This may have implications for use cases that require pristine or fully anonymized text generation, although Anthropic has not yet detailed any opt-out mechanisms. The company has said that watermarking is applied at the model level, so it cannot be disabled by a user or administrator.

There are also potential consequences for AI-assisted writing practices. Many professionals use AI tools for drafting, brainstorming, or editing, and they may not want the final text to be flagged as machine-generated. Anthropic acknowledges that the watermark may persist through some editing, but the company has not quantified how much rewriting is needed to remove it. This uncertainty could become a point of friction for users who rely on AI-generated content in creative or journalistic workflows.

Challenges and Limitations of Text Watermarking

Text watermarking remains a technically difficult problem. Unlike image or video content, text has relatively little redundancy, and any alteration can destroy subtle statistical signals. Paraphrasing, translation, or even simple grammar changes can potentially remove a watermark. Researchers have proposed various methods, including synonym substitution patterns, punctuation adjustments, and controlled sampling during generation. However, no approach is foolproof, and determined users can often find ways to circumvent the marker.

Another concern is the potential for false positives. If watermarking is applied to all AI-generated text, there is a risk that human-written text might be misidentified as AI-generated due to stylistic similarities. This is especially relevant in areas like academic writing, where standardized structures and formal language are common. Companies like Anthropic will need to calibrate their detection algorithms carefully to minimize incorrect attributions.

Privacy advocates have also raised questions about the implications of watermarking for anonymous speech. If all AI-generated content is traceable, it could make it easier for governments or companies to track who is generating what. Anthropic has not disclosed whether the watermark contains any information about the user, the time of generation, or the specific model version. The company says the watermark is part of the text, but it is unclear whether it encodes user identity or just a generic marker.

Looking Ahead

Anthropic’s decision to watermark text is a significant step toward compliance with the EU AI Act and reflects a broader industry shift toward transparency. As AI models become more capable and more widely used, the ability to distinguish AI-generated content from human-authored content will become increasingly important. The company’s support page promises that watermarking will be present in all Claude products, and that older models will also be covered, but the technical details and enforcement mechanisms are still evolving.

The industry will be watching to see how effective Anthropic’s watermarking proves in practice. The true test will come when the watermarked text is circulated across the internet, edited, reformatted, and subjected to paraphrasing tools. If the watermark survives a reasonable amount of editing, it could set a new standard for AI transparency. If it fails, it will underscore the difficulty of regulating generative AI through technical means alone.

For the time being, Anthropic has joined a growing list of companies that are taking proactive steps to label AI-generated content. The combination of regulatory pressure, public concern, and competitive dynamics is pushing the entire industry toward greater openness. Whether these efforts fully solve the problem of AI-generated misinformation remains to be seen, but they represent a meaningful beginning.

Source: TechCrunch News

Key facts at a glance

• Fundraising: $45 million Series B, led by Peak XV Partners with participation from GV and Y Combinator.
• Valuation: $550 million, up from $60 million in less than a year.
• Total funding: $58.5 million.
• Founded: 2024.
• CEO: Aditya Jayaprakash.
• Customers: More than 5,000, including Mercury, Supabase, Clerk, Ashby, and Expensify.
• Products: Cloud-based CI platform and Codesmith, an AI coding agent.
• Revenue: Reached $10 million annualized run rate with 10 employees; now in the tens of millions with about 30 employees.

A rapid valuation surge

The new round marks one of the fastest valuation jumps in the AI developer tools space. Blacksmith went from a $60 million post-money valuation to a $550 million valuation in under a year, reflecting surging demand for tools that help software teams manage the flood of AI-generated code. The company's customer base grew from roughly 700 to more than 5,000 in the same period, a clear signal that the market is embracing its approach.

The funding environment for AI infrastructure and developer tools has been intensely competitive, with investors pouring money into startups that address bottlenecks created by generative AI. Peak XV Partners, which led the round, is one of the largest venture firms focused on technology startups in Asia and has backed several high-growth SaaS companies. GV, the venture capital arm of Alphabet, and Y Combinator, the renowned startup accelerator, have both supported Blacksmith since its early days.

The speed of the valuation increase also underscores how quickly the developer tools landscape is being reshaped by AI. Software teams are generating code at a pace that was impossible just a few years ago, but the ability to verify that code remains a major constraint. Blacksmith sits at the intersection of two rapidly evolving categories: continuous integration and automated testing, and AI-powered coding assistants.

Building on a simple but critical pain point

Blacksmith was founded in 2024 with a focus on helping companies build, test, and verify software before it reaches production. The company started as a cloud provider for continuous integration, or CI, workloads. CI is a software engineering practice where developers frequently merge their code changes into a shared repository, and automated builds and tests are run every time a change is made. This helps catch bugs early, but it requires significant computing power and careful orchestration.

Many developer teams have struggled with the cost and complexity of running CI pipelines, especially as projects grow and AI tools accelerate the volume of code changes. Blacksmith's platform was designed to make CI faster and more affordable, giving developers quick feedback on whether their code is safe to merge. That foundation proved valuable, but the company quickly realized that the bigger challenge was validating code written by AI.

The rise of AI coding tools such as Cursor, OpenAI's Codex, and Anthropic's Claude Code has made it significantly easier for software teams to generate code. Instead of manually writing every function, developers can describe what they want and let an AI assistant produce the implementation. While that boosts productivity, it does not guarantee that the generated code is correct, secure, or free of bugs.

“Validating code is still a bottleneck, and it's an even bigger bottleneck because people are writing even more,” said Aditya Jayaprakash, co-founder and CEO of Blacksmith. His point reflects a growing concern across the industry: AI accelerates the creation of code, but the teams responsible for reviewing, testing, and deploying that code are still limited by the same time and compute resources as before.

Codesmith and the shift toward AI-assisted testing

To address this bottleneck, Blacksmith has broadened its platform with Codesmith, an AI coding agent that can automatically fix failed code checks. Rather than simply flagging an error and asking a developer to investigate, Codesmith analyzes the failure and proposes a remediation. This moves the company beyond infrastructure and into the AI layer of the software development lifecycle.

Codesmith is designed to integrate with the existing testing workflow. When a code change triggers a build or test failure, Codesmith reviews the logs, identifies the root cause, and generates a fix. The developer can then review the suggested patch and decide whether to accept it. This type of human-in-the-loop approach is becoming common in AI-assisted development, as it balances automation with oversight.

The introduction of Codesmith also reflects a broader industry trend. AI coding agents are no longer just about generating code; they are increasingly being used to review code, write tests, fix vulnerabilities, and automate maintenance tasks. Blacksmith intends to compete in this broader category, aiming to provide a complete suite of tools that help developers write, validate, and merge software faster.

Customer growth and revenue momentum

Blacksmith now counts more than 5,000 customers, including well-known technology companies such as Mercury, Supabase, Clerk, Ashby, and Expensify. That is a striking increase from the 700-plus customers it had less than a year ago. The company has not disclosed a precise updated annualized revenue run rate, but Jayaprakash said revenue has grown to “tens of millions of dollars” and that some of its largest customers now spend more than $1 million a year on the platform.

The company reached a $10 million annualized revenue run rate with just 10 employees, a sign of strong product-market fit and efficient growth. It has since expanded its team to about 30 people, still a remarkably small headcount for a company at this scale. This lean operation is possible because the product is largely automated and delivered as a cloud service.

Customer concentration in the developer tools space can be volatile, as engineering teams often switch tools based on price, performance, and reliability. But Blacksmith's rapid customer growth suggests that it has found a niche that larger competitors have not fully addressed. Its focus on speed and affordability appears to be resonating, especially with startups and mid-sized companies that need fast CI without enormous cloud bills.

A crowded and competitive market

Blacksmith operates in a crowded and fiercely competitive market. Its key competitors include GitHub Actions, which is integrated into the world's largest code-hosting platform; Cursor Automations, which come from one of the most popular AI code editors; and validation capabilities built directly into AI models like Codex and Claude Code. In addition, major cloud providers including Amazon Web Services, Microsoft Azure, and Google Cloud all offer AI code-testing and CI-related services.

Each of these competitors has advantages. GitHub Actions benefits from its deep integration with GitHub repositories. Cursor has a loyal user base among developers who use its AI-powered editor. The cloud providers can bundle testing with their broader cloud offerings, making it convenient for enterprises to stay within one ecosystem. Startups like Blacksmith must therefore compete on more than just features.

Jayaprakash said his startup is competing on speed and affordability. In the world of CI, speed is a productivity metric: every minute a developer waits for a build or test result is a minute they cannot spend on writing new code or fixing issues. Affordability is equally important, as CI costs can spiral out of control when teams have long build queues and expensive compute instances. Blacksmith aims to deliver fast, cost-effective testing without requiring customers to manage their own infrastructure.

The competitive pressure has not slowed Blacksmith's growth, but it does raise questions about long-term differentiation. The company's bet is that a dedicated platform focused on validation will outperform general-purpose coding assistants and cloud-native CI tools. By combining traditional CI infrastructure with an AI agent that can automatically fix failures, Blacksmith hopes to create a workflow that developers will not want to leave.

Future plans

Looking ahead, Blacksmith plans to expand into a broader suite of coding tools. The company wants to cover more of the software development lifecycle, from initial code generation to testing, validation, and deployment. Jayaprakash has not revealed exactly which products will come next, but the direction is clear: Blacksmith aims to be more than a CI provider or a one-trick AI assistant.

The company is likely to invest heavily in Codesmith, integrating it more deeply into the developer workflow and expanding its ability to handle complex codebases. It may also add features around security scanning, code review, and performance testing, as these are natural complements to the existing platform. Given the rapid pace of AI model improvements, Blacksmith will need to continuously adapt its AI agent to work with new models, programming languages, and frameworks.

Blacksmith's rapid rise is a reflection of the broader AI-driven transformation in software development. As AI coding tools become more powerful, the need for robust testing and validation will only grow. Blacksmith has positioned itself to address that need, and its new funding gives it the resources to pursue a larger share of the developer tools market. With a valuation of $550 million and a total funding haul of $58.5 million, the company is now well capitalized to compete against much larger rivals.

The company's trajectory also illustrates how quickly startups can scale in the AI era when they address a critical pain point. In less than a year, Blacksmith grew from a small CI cloud provider into a significant player in the AI code-testing space, with thousands of customers and tens of millions in revenue. The next year will likely determine whether it can maintain that momentum and fend off competition from the biggest names in technology.

Source: TechCrunch News

The online discussion platform serves as a central destination for technology enthusiasts, PC builders, gamers, and professionals seeking advice on hardware, software, and consumer electronics. With hundreds of thousands of threads and millions of messages, the community offers a vast knowledge base. The forum is currently addressing a temporary front-end styling issue, but moderators have confirmed that user data and functionality remain unaffected.

Hardware and Technology Discussions

The Hardware and Technology section is the core of the community. It includes multiple specialized boards covering processors and overclocking, motherboards, graphics cards, memory and storage, displays, power supplies, cooling, laptops, networking, Apple computers, peripherals, and general computer building. Each board features active discussions, troubleshooting help, and product recommendations.

In the CPUs and Overclocking board, members exchange tips on maximizing processor performance. The board hosts over 78,000 threads and 1.4 million messages. Recent conversations include speculation about next-generation CPU architectures. The Motherboards board, with 67,000 threads, focuses on compatibility, BIOS settings, and system stability. A recent thread discusses wake-from-sleep malfunctions and BIOS ERP settings.

Graphics Cards is one of the most active areas, with 96,000 threads and 1.6 million messages. It covers GPU architecture, gaming performance, and driver issues. The board is divided into sub-forums for AMD and Nvidia, allowing brand-specific discussions. A recent topic highlights advancements in gaming graphics and adjacent software.

Memory and Storage covers DDR3, DDR4, SSDs, HDDs, flash memory, and USB drives. With 22,600 threads, members discuss RAM pricing trends and storage reliability. Displays, with 11,200 threads, focuses on monitors and display devices, including troubleshooting black screen issues after power outages. Power Supplies has 6,000 threads and addresses power requirements for high-end systems. Cases and Cooling, with 46,900 threads, helps users choose cases and optimize thermal performance.

Laptops and mobile slates are covered in a dedicated board with 24,000 threads, including hardware issues and purchase advice. Networking, with 54,600 threads, assists with LAN, WAN, routers, and server configuration. Apple Laptops and Desktops includes MacBook and iMac discussions, with 895 threads. Peripherals and Components covers keyboards, mice, headsets, and printers, with 18,500 threads. Computer Building, one of the largest areas, has 314,000 threads and 2.5 million messages, with sub-forums for pre-built desktops, barebones systems, Raspberry Pi, and home theater PCs.

Consumer Electronics

Beyond PC hardware, the forum includes Consumer Electronics categories. Digital Cameras and Video hosts 5,600 threads about photography and videography equipment, including mirrorless cameras and lenses. Console Gaming features 11,900 threads covering retro and current consoles, with the official PlayStation 5 thread among the most popular. Mobile Devices, with 17,200 threads, discusses smartphones, tablets, and cellular providers. Audio Components, with 10,900 threads, focuses on speakers, headphones, amplifiers, and audiophile setups. TVs, with 4,100 threads, covers display technology, home theater configuration, and broadcast standards. Recent topics include ATSC 3.0 patent concerns and personal hi-fi system longevity.

Software and Operating Systems

The Software section addresses Windows, Apple, open source, operating systems, programming, PC games, distributed computing, and security. Windows discussions, numbering 101,000 threads, include application performance and system resource usage. Apple discussions cover macOS, iOS, and Mac hardware. Open Source and Operating Systems boards, with 3,100 and 73,900 threads respectively, span Linux, Unix, Windows, and cross-platform tools. Programming has 6,800 threads for coding questions. PC Games, with 20,200 threads and 678,000 messages, is a bustling hub for game recommendations and free game announcements. Distributed Computing, with 40,900 threads, supports projects like BOINC, PrimeGrid, and Folding@Home. Security, with 3,700 threads, covers vulnerabilities, backdoors, and best practices.

Shopping and Deals

The Shopping section helps members find discounts and product promotions. Hot Deals and Giveaways has 125,000 threads and 1.7 million messages. Users frequently post price drops from major retailers, such as a recent Intel Core Ultra 7 processor deal. Black Friday discussions are also present, with 92 threads and 808 messages, focusing on seasonal sales and events like Prime Day.

Social and Community Areas

The Social section provides spaces for off-topic conversation and special interests. OT Discussion Club has 794,000 threads and 18.3 million messages, covering everything from world events to casual banter. Politics and News, with 84,200 threads, encourages serious discussion of current events. Ask a Technical Professional is designed for in-depth PC architecture and technology discussions. The Garage covers automotive topics, while Health and Fitness and Home and Garden offer lifestyle advice. A recent health thread encourages preventive screenings, and a home improvement thread compares concrete slab and pier construction methods.

Forum Issues and Feedback

The Forum Issues and Feedback board, with 16,400 threads, allows users to report technical problems and provide feedback. The community is currently aware of a layout and styling issue affecting page display. According to a notice, the problem has been identified and a fix is being deployed. The notice emphasizes that there is no impact to user data or functionality, and the issue is strictly front-end display.

Community Engagement and Knowledge Sharing

The forum's strength lies in its active membership and the depth of knowledge shared across categories. With millions of messages, users can find answers to obscure hardware questions, troubleshooting guides, and carefully curated deal listings. The structured sub-forums ensure that conversations stay organized, from GPU brand-specific threads to single-board computer projects. This environment fosters both beginner-friendly help and advanced technical debate.

Recent activity illustrates the diversity of discussions. Within minutes of any given day, members post about gaming graphics innovations, World Cup events, distributed computing challenges, and political topics. This constant stream of content makes the forum a dynamic resource. Whether a user is looking for advice on a BIOS setting, a laptop repair, or a new power supply, the community offers practical solutions grounded in real-world experience.

The forum also serves as an archive of technology trends. Threads from previous years reveal how hardware has evolved, from older DDR3 memory to current high-wattage power supplies. This historical context is valuable for researchers, enthusiasts, and buyers trying to understand the trajectory of the industry. The inclusion of dedicated boards for Apple products and open-source software reflects the broad range of interests represented in the community.

Moderation and rules play a crucial role in maintaining quality. Several boards explicitly prohibit flame wars and off-topic posts, creating a welcoming environment for technical help. The Ask a Technical Professional board, for example, requires serious, on-topic discussions about PC architecture. Such guidelines help keep the signal-to-noise ratio low, especially in high-traffic areas.

As the front-end display issue is resolved, the community will continue to operate as a hub for technology news and discussion. The staff's transparency about the ongoing maintenance builds trust among members. The forum's longevity is a testament to its committed user base, which has contributed millions of posts over the years. New members are encouraged to register, install the web app, and participate in the conversations that define this vibrant online community.

Source: AnandTech Forums: Technology, Hardware, Software, and Deals News

The Rise of Hybrid and Electric Vehicles

Hybrid vehicles, which combine an internal combustion engine with an electric motor and battery, have been available for decades. The Toyota Prius, launched in the late 1990s, demonstrated that fuel-efficient hybrids could appeal to mass-market buyers. Since then, the market has expanded dramatically, with nearly every major automaker offering at least one hybrid or plug-in hybrid model.

Fully electric vehicles, powered solely by batteries, have seen even more explosive growth in recent years. Global EV sales surpassed 10 million units in 2022, accounting for roughly 14% of all new car sales, according to industry estimates. By 2023, that share had grown further, with markets like Norway, China, and parts of Europe leading the way. Analysts predict that by 2030, electric vehicles could represent more than half of all new car sales worldwide.

Environmental Benefits and Challenges

The primary driver behind the push for electrification is the need to reduce greenhouse gas emissions and improve urban air quality. Transportation is responsible for a significant share of global carbon dioxide emissions, and electric vehicles offer a pathway to decarbonize personal mobility. Even when accounting for electricity generation, EVs typically produce fewer lifetime emissions than comparable gasoline-powered cars.

However, the environmental benefits depend heavily on how the electricity is generated. In regions with coal-heavy grids, the emissions savings are smaller, although still positive in most cases. As renewable energy sources such as wind and solar expand, the carbon footprint of EVs will continue to decline.

There are also concerns about battery production and disposal. Mining lithium, cobalt, and nickel has environmental and social impacts, and battery recycling is still developing. Automakers and governments are investing in sustainable sourcing and closed-loop recycling systems to address these issues. Advances in battery chemistry, such as lithium iron phosphate and solid-state batteries, promise to reduce reliance on rare and conflict-prone materials.

Battery Technology and Range

Battery technology is the heart of the electric vehicle revolution. Early EVs had limited range and long charging times, making them impractical for many drivers. Today, modern electric cars routinely offer ranges of 250 to 400 miles on a single charge, with some premium models exceeding 500 miles.

Lithium-ion batteries dominate the market, and their energy density has improved steadily while costs have fallen dramatically. The average cost of a lithium-ion battery pack dropped from over $1,100 per kilowatt-hour in 2010 to around $130 per kilowatt-hour by 2023. This price decline has made EVs more affordable and closer to price parity with gasoline cars.

Solid-state batteries are considered the next major breakthrough. They promise higher energy density, faster charging, and improved safety by replacing the liquid electrolyte with a solid material. Toyota, Volkswagen, and other manufacturers are racing to commercialize solid-state technology, though production scale-up remains challenging. In the meantime, ultra-fast charging networks capable of delivering 350 kW or more are reducing charging times to 15–20 minutes for many models.

Charging Infrastructure Expansion

One of the biggest barriers to EV adoption has been the availability and reliability of public charging stations. Range anxiety—the fear of running out of battery with no place to charge—remains a concern for potential buyers. Governments and private companies are investing heavily in charging infrastructure to ease these worries.

In the United States, the Bipartisan Infrastructure Law allocated $7.5 billion to build a national network of EV chargers, with a focus on highway corridors and underserved communities. The European Union has set targets for millions of public charging points by 2030. China already leads the world in public charging infrastructure, with millions of chargers installed across the country.

Home charging is also essential. Many EV owners charge overnight at home using a Level 2 wall box, which adds roughly 25 to 30 miles of range per hour. For multi-unit housing and urban dwellers, curbside charging and workplace charging are being expanded. Wireless inductive charging is emerging as a convenient option, though it is not yet widely deployed.

Government Incentives and Policies

Government policies have played a critical role in accelerating the transition to electric mobility. Purchase incentives, tax credits, and rebates lower the upfront cost of EVs, making them more accessible to consumers. The U.S. federal tax credit of up to $7,500 for qualifying EVs has been a major stimulus, though the rules have been updated to encourage domestic battery manufacturing and sourcing.

Several countries have announced plans to phase out the sale of new internal combustion engine vehicles in the coming decades. The European Union has approved a ban on new gasoline and diesel car sales by 2035. Norway, a global leader in EV adoption, aims to sell only zero-emission vehicles by 2025. China has set a target of 50% of new car sales being electrified by 2035, with strong support for plug-in hybrids and battery-electric vehicles.

Regulations on emissions and fuel economy are also pushing automakers to electrify their lineups. Fleet average emission standards are becoming stricter, and companies that fall short face hefty fines. As a result, automakers are investing billions of dollars in electric vehicle development, battery production, and supply chain partnerships.

Economic and Market Dynamics

The shift to electric vehicles is reshaping the global automotive economy. Traditional automakers such as Ford, General Motors, Volkswagen, and Toyota are transitioning their vehicle portfolios, while new entrants like Tesla, BYD, Rivian, and NIO are challenging established players. Tesla remains the world's most valuable automaker, but BYD surpassed it in sales volume for electric vehicles in some quarters, reflecting the rapid growth of the Chinese market.

Job creation is another important dimension. EV manufacturing requires fewer parts than traditional vehicles—an electric drivetrain has about 20 moving parts compared with hundreds in an internal combustion engine. This simplicity reduces maintenance costs for owners but can also reduce the number of jobs in engine and transmission production. At the same time, new jobs are being created in battery manufacturing, software development, and charging infrastructure installation.

The used EV market is beginning to mature. As early models come off lease and newer vehicles with longer range enter the market, used electric cars are becoming more affordable. Battery health and remaining capacity are key considerations for used EV buyers, and some manufacturers offer warranties of eight years or 100,000 miles for battery packs.

Types of Hybrid and Electric Vehicles

Consumers face a variety of electrified powertrain options, each with different trade-offs. Conventional hybrids (HEVs) use a gasoline engine paired with an electric motor and a small battery; they cannot be plugged in and rely on regenerative braking to recharge. They offer improved fuel economy without changing driver habits, but they still emit tailpipe emissions.

Plug-in hybrids (PHEVs) have larger batteries that can be charged from an outlet, providing a limited all-electric range—typically 20 to 50 miles—before the gasoline engine kicks in. This allows many daily trips to be completed entirely on electricity while retaining the ability to take long journeys without relying on charging stations. PHEVs are often seen as a bridge between conventional cars and full EVs.

Battery electric vehicles (BEVs) rely solely on electricity and produce zero tailpipe emissions. They offer the lowest running costs and simplest maintenance but require access to reliable charging. Fuel cell electric vehicles (FCEVs), which use hydrogen to generate electricity, are another zero-emission option, though their infrastructure remains sparse.

Consumer Adoption and Challenges

Despite the rapid growth, several challenges remain. The upfront cost of electric vehicles is still higher than comparable gasoline models, although total cost of ownership can be lower when factoring in fuel and maintenance savings. Charging times, even with fast chargers, are longer than filling a gas tank, which can be inconvenient on road trips.

Cold weather can also reduce EV range significantly, as batteries operate less efficiently at low temperatures. Automakers are addressing this with heat pump systems and improved battery thermal management. Additionally, the availability of public charging in rural areas and low-income neighborhoods lags behind urban centers, raising questions about equity in the transition to electric mobility.

Consumer awareness is improving, but many prospective buyers are still unfamiliar with EV ownership. Dealerships and automakers are investing in education and test-drive experiences to help consumers overcome hesitation. Automakers are also expanding their electric model lineups to include SUVs, pickup trucks, and commercial vans, which dominate new vehicle sales in many markets.

The Future of Hybrid and Electric Vehicles

Looking ahead, the momentum behind electrification is unlikely to slow. Technological innovation continues to drive down costs and improve performance. Autonomous driving features, over-the-air software updates, and connectivity are becoming standard in many electric vehicles, enhancing their appeal. The integration of electric vehicles with smart grids could allow vehicle-to-grid applications, where car batteries store excess renewable energy and feed it back during peak demand.

Automakers are committing to electric-only futures. General Motors plans to offer only electric light-duty vehicles by 2035, and Ford has announced similar ambitions in Europe. Stellantis, BMW, Mercedes-Benz, and Volvo have all set ambitious electrification targets. Meanwhile, battery manufacturers are expanding gigafactory capacity around the world to meet expected demand, and recycling technologies are evolving to create a circular supply chain.

The shift to hybrid and electric vehicles is not just a change in powertrain technology; it represents a fundamental change in how people think about transportation. It offers the prospect of cleaner air, reduced dependence on fossil fuels, and a more sustainable transportation system. The road ahead is still long, but the destination is increasingly clear.

Source: TechRadar News

Why Virginia is at the center of the AI data center boom

Virginia, and particularly Northern Virginia, is one of the largest data center markets in the world. The region is often called "Data Center Alley" because of the enormous concentration of server farms located there. For years, the state has attracted cloud computing companies and internet giants with its business-friendly tax policies, robust infrastructure, and proximity to the nation's capital.

The rapid expansion of artificial intelligence has added new fuel to that growth. AI applications require vastly more computing power than traditional cloud services, and that computing power must live somewhere. Data centers are being built, expanded, and powered up at an unprecedented pace. This has created a new surge in electricity demand, and utilities have been quick to point to that demand when asking for permission to raise rates.

Utilities argue that they need billions of dollars in new transmission lines, substations, backup systems, and power generation capacity to meet the needs of these massive facilities. In many cases, they have proposed rate increases that would apply broadly across their customer base, including to homes, schools, hospitals, and small businesses.

The problem with spreading costs among all customers

The challenge for regulators is a delicate one. Data centers bring jobs, tax revenue, and economic development. But they also create an enormous strain on the electric grid. If utilities recover the costs of building new infrastructure only by raising rates for everyone, then families and small businesses could end up paying for upgrades that are mainly serving large technology companies.

Consumer advocates have warned that such cross-subsidization is unfair. They argue that the biggest users of electricity should bear a proportionate share of the network modernization costs. When data centers triple or quadruple their power usage in a short period, the grid must be expanded rapidly. And if that expansion is financed through across-the-board tariff increases, household customers face what some have described as a hidden tax on their monthly utility bills.

This concern has become acute in Virginia, where electricity demand is projected to grow at rates not seen in decades. Some forecasts suggest that the state's grid will need to double or even triple its capacity over the next twenty years, largely because of data centers and AI workloads. Without careful regulation, the burden of that transformation could fall heavily on ordinary residents.

What Virginia regulators have decided

The recent action by Virginia regulators changes the equation. Rather than allowing electricity firms to simply fold data center related costs into general rate increases, the state has directed utilities to treat those costs separately and ensure that the pricing reflects the specific demands of these facilities. The goal is to prevent artificial intelligence data centers from being subsidized by residential and small business customers.

This means utilities will need to design rates that more accurately match the cost of serving large commercial power users. For example, data centers that require new transmission infrastructure or dedicated substations could be required to pay a larger share of those capital expenses directly. They may also face demand charges that reflect the strain they place on the grid at peak hours.

According to preliminary estimates cited in the announcement, the crackdown could save residential customers hundreds of millions of dollars that would otherwise have been added to their bills over the next several years. That range is difficult to quantify precisely because utility rate cases are complex and many factors remain in play. But the direction is clear: ordinary ratepayers should not be the ones bankrolling the AI boom.

How Northern Virginia became "Data Center Alley"

Northern Virginia's dominance in the data center world is no accident. In the late 1990s and early 2000s, the Washington, D.C. suburbs became a natural hub for internet infrastructure because of robust fiber-optic networks, abundant skilled workers, and close proximity to federal agencies that were early adopters of cloud computing. Loudoun County, in particular, has become the anchor of the digital economy, with millions of square feet of server space and more planned every year.

The growth of commercial cloud platforms like Amazon Web Services helped cement the region's status. Over time, the county learned how to attract and manage these developments, turning itself into a model for other parts of the country. But the physical footprint of data centers has expanded far beyond what was once imagined, and with it has come a corresponding surge in energy demand.

The electricity infrastructure that was seen as a great advantage in the early days is now under severe pressure. Local transmission lines are operating close to capacity, and utility companies say they need major upgrades just to ensure reliability. The question has been who should pay for those upgrades, and that is where the controversy lies.

The debate over who pays for grid upgrades

The utility business model is built around recovering costs through rates set by regulators. When a company spends billions of dollars on a new power plant or transmission line, it needs a predictable way to earn that money back. If the investment serves only a handful of data centers, charging only those customers is the fairest and most efficient solution.

However, utilities have often argued that grid upgrades benefit all customers by providing improved reliability and system resilience. That argument has some merit, but consumer groups say it can be a convenient excuse for saddling everyone with higher costs. The new regulatory stance in Virginia narrows the path for that kind of reasoning, at least where dedicated infrastructure for AI facilities is concerned.

It also creates an incentive for utilities to be more rigorous about planning. Instead of simply building new capacity and socializing the costs, they will need to demonstrate which customers are requesting the capacity and how those customers will pay for it. That could lead to more accurate pricing and better long-term resource allocation.

Potential savings for families and small businesses

For the average home, electricity bills are already a significant share of monthly expenses. Air conditioning in Virginia's humid summers and electric heating in the winter can push costs even higher. An unexpected rate increase driven by data center construction would be difficult for many families to absorb, especially those already struggling with inflation and other rising costs.

The savings from the crackdown are expected to be widespread. While individual households may not see a direct refund, the avoidance of new charges could mean the difference between a modest increase in their annual utility costs and a dramatic one. For small businesses, the impact is even more pronounced because they often operate on thinner margins.

Regulators have also signaled that they will scrutinize any future rate requests more closely, looking for signs that residential customers are being unfairly asked to support industrial growth. This could lead to a long-term change in the way utility rates are structured in the state.

The response from the technology and utility industries

Unsurprisingly, the decisions from Virginia's regulators have sparked mixed reactions in the industry. Technology companies that operate data centers say they are committed to working with utilities to ensure reliable power, and many have already signed contracts that tie their growth to new renewable energy projects. Some have expressed concern that new regulations could slow down the permitting and construction of essential grid infrastructure.

Utility companies have also pushed back, warning that costs for data center customers could eventually be felt by everyone if projects are delayed or canceled. They argue that building infrastructure at the speed required by AI will require some form of shared investment, at least initially, to maintain grid stability for all users.

But Virginia regulators have held firm, noting that the state has enough tools to accelerate infrastructure development without compromising consumer protections. They can use special tariff structures, priority review processes, and other mechanisms to keep projects moving forward while ensuring that the financial impact is allocated fairly.

How other states are watching

The outcome of Virginia's experiment is likely to influence energy policy in other parts of the country. States like Texas, Georgia, Ohio, and California are also experiencing major data center growth, and they face the same basic question: who pays for the grid upgrades? Each state has a different regulatory structure, but the core tension between economic development and consumer protection is universal.

Early signs suggest that the Virginia approach could become a template. If data centers continue to be built at a strong pace while household bills remain under control, other states may follow suit. Conversely, if the regulations lead to prolonged legal battles and weaker grid investment, the opposite lesson could be drawn. Either way, the stakes are high because AI energy demand is not expected to fade anytime soon.

The road ahead for AI energy policy

As artificial intelligence becomes more integrated into daily life, from search engines to customer service and medical research, the demand for computing power will only accelerate. Managing that growth responsibly is one of the defining energy policy challenges of the coming decade.

Virginia's crackdown on electricity rate increases is an early warning to utilities and technology companies that ordinary consumers cannot be treated as a backstop for corporate ambitions. It is also a reminder that the benefits of the AI revolution should not come at the expense of the families and small businesses that form the backbone of the local economy.

In the months ahead, utilities will likely seek to negotiate a more flexible path forward, but the regulatory principle has been established: companies that create the demand should also carry the costs. That principle, if sustained, will save Virginians hundreds of millions of dollars and set an example for power grids everywhere.

Source: TechRadar News

The situation in Gilroy reflects a larger pattern emerging across California and the United States, as tech giants rush to build the massive computing infrastructure required for artificial intelligence. Data centers consume enormous amounts of electricity and water, generate noise and traffic, and often sit on land that communities have other plans for. In this case, residents say they were caught off guard when construction crews appeared at the site, and they are now demanding answers from both Amazon and local officials.

What Are the 45-Year-Old Rules?

The specific rules Amazon is reportedly using are not new. They were created in an era when large infrastructure projects, including power plants and transmission lines, were often fast-tracked to meet the state's growing energy demands. The legal framework is understood to give certain utility or energy-related projects a streamlined path through the permit process, sometimes overriding local zoning and planning restrictions. Environmental review requirements may still apply, but the scope of that review can be narrower than for typical commercial developments.

By framing the data center as an extension of critical energy infrastructure, Amazon may be able to avoid lengthy public hearings and some discretionary approvals. That interpretation has infuriated residents who thought they would have a chance to weigh in on a project that will consume significant local resources. Legal experts say the rules were designed decades ago for a very different kind of industrial development, and applying them to a modern AI data center stretches their original intent.

The age of the rules is central to the dispute. They were written long before cloud computing, machine learning, or the explosive growth of data centers. Critics argue that using 45-year-old regulations to authorize a contemporary AI facility is a misuse of the law. Supporters of the project, on the other hand, say the rules exist precisely to prevent bureaucratic delays from stalling essential infrastructure. The tension between those two views is now playing out in the streets of Gilroy.

Why Gilroy?

Gilroy is a city of about 60,000 people located in Santa Clara County, at the southern end of Silicon Valley. It is widely known as the Garlic Capital of the World and has historically been an agricultural community, with vast fields of garlic, strawberries, and other crops surrounding the urban core. The city has grown in recent decades as housing costs in San Jose and San Francisco have pushed people farther south, but it still retains a distinctly rural character compared to its tech-heavy neighbors.

That agricultural heritage is one reason residents are especially sensitive to the arrival of a large data center. Land that may have supported farming for generations is now being converted to industrial use, and many fear the character of the community will change permanently. Data centers do not employ large numbers of people, so the economic benefits are often limited to property tax revenue and a handful of maintenance jobs. For local residents, the burden of noise, traffic, and strain on the electrical grid can outweigh those benefits.

Gilroy is also attractive to data center developers because of its proximity to major fiber optic lines, its relatively affordable land, and its location in a county that already houses many technology infrastructure facilities. The city has good access to highways and a mild climate that can reduce cooling costs. But those advantages are exactly what makes the town vulnerable to development pressures from the fast-growing AI industry.

The AI Data Center Boom

The new Gilroy facility is part of a much larger trend. Artificial intelligence models require vast amounts of computational power, and tech companies are building data centers at an unprecedented pace to train and run those models. Unlike traditional data centers, AI facilities are often designed to operate at very high densities, with advanced cooling systems and massive power supplies. They can consume as much electricity as a small city, and they run around the clock.

In California, where energy is already expensive and the grid is under stress during hot summer months, the arrival of multiple AI data centers has become a contentious issue. Utility companies have warned that the demand from data centers could require new power plants and transmission lines, along with major upgrades to the local distribution network. Some of those costs may ultimately be passed on to ratepayers. Community groups have begun to ask whether AI infrastructure should be prioritized over housing, schools, and other public services.

Amazon is one of the largest players in the cloud computing and AI infrastructure space. Its AWS division operates data centers around the world and has committed to spending heavily on new facilities to support AI services. The company has often preferred locations with easy access to renewable energy and favorable tax treatment. In California, however, the approval process for new construction can be lengthy and litigious. Developers may try to find regulatory shortcuts, and the use of older laws is one way to overcome obstacles.

Local Concerns and Anger

Residents in Gilroy have described their frustration in public meetings and social media posts. Many say they learned about the project only after bulldozers and cranes appeared on site. Others worry about the impact on groundwater supplies, since data centers use significant amounts of water for cooling. There are also concerns about diesel generators, air quality, and the visual impact of a large industrial building in a predominantly agricultural area.

The lack of meaningful public participation is perhaps the most inflammatory issue. Typically, a large development project in California would require environmental review under the California Environmental Quality Act, also known as CEQA. That review process includes an opportunity for public comment and often leads to modifications of the project to address community concerns. But if the project qualifies for an exemption or falls under a different regulatory threshold, those opportunities can be limited or eliminated altogether.

Residents have demanded that local officials investigate whether the project should be halted until a full review can be conducted. They have also asked the city to clarify what legal basis Amazon is using to move forward. Some have hired attorneys and environmental groups have begun to look into potential challenges. But with construction already underway, the burden is on opponents to prove that the project is not entitled to the expedited treatment it has received.

A Broader Regulatory Debate

The Gilroy dispute is likely to have implications beyond Santa Clara County. As the AI industry expands, more communities will face similar decisions about how to accommodate data centers. The question is not only whether these facilities should be built, but who gets to decide. Old rules that were designed to accelerate infrastructure development are now being tested in new contexts, and courts may eventually be asked to determine whether they apply to a technology that did not exist when the laws were written.

State lawmakers are also paying attention. Several bills have been introduced in California in recent years to reform the data center approval process, increase transparency, and require more robust environmental assessments. Industry groups have pushed back, arguing that excessive regulation will drive data center investment to other states. That argument carries weight in a time when every state wants the jobs and tax revenue associated with AI. But local residents may not see the benefit if their quality of life suffers while the profits go to distant corporations.

Amazon, for its part, has emphasized the importance of data centers for the digital economy and has said it is committed to sustainability. The company has pledged to match its electricity use with renewable energy and has made ambitious climate commitments. However, those global promises do little to address the specific complaints of people who live next to a construction site. A company can be a good corporate citizen in the abstract while still creating noise, dust, and congestion in a particular neighborhood.

The situation in Gilroy is a reminder that the AI era is not only about exciting new products and services. It is also about physical infrastructure, land use, and the power of governments to regulate that infrastructure. The companies building the behind-the-scenes facilities for artificial intelligence have learned that the biggest obstacles are often not technological, but legal and political. Using old rules to bypass those obstacles may work in the short term, but it can generate a backlash that leads to new rules and stricter oversight.

As the Gilroy facility continues to rise from the ground, the anger among residents shows no signs of fading. Community meetings have drawn large crowds, and local leaders have promised to look into the issue. The company has remained relatively quiet, apparently confident in its legal position. Yet the dispute has already become a symbol of the collision between high-tech ambitions and local democracy. What happens in the coming weeks will likely shape how other data centers are approved elsewhere in the state, and perhaps across the country.

Source: TechRadar News

For 2026, major tech manufacturers have doubled down on accuracy, durability, and user experience. Advanced algorithms now provide deeper insights into recovery, sleep quality, and even stress levels. Meanwhile, integration with smart home devices and contactless payment systems has become seamless. Below, we explore the market-leading devices, their standout features, and what to look for before making a purchase.

Why Smartwatches Have Become Essential

Once considered a luxury gadget, the smartwatch has become a daily essential for many. The ability to glance at your wrist to check messages, control music, or track your heart rate during a workout saves time and keeps you present. In 2026, these devices are also increasingly central to health management. Many models now offer electrocardiograms (ECG), blood oxygen monitoring, and even temperature sensing. For people with chronic conditions, these tools can provide valuable data to share with healthcare providers.

Furthermore, the social aspect of fitness tracking has grown. Wearable tech now includes advanced community features, allowing friends to compete in challenges, share achievements, and motivate each other. This gamification pushes users to become more active, transforming exercise into a rewarding experience. As medical research increasingly validates consumer wearable data, expect these devices to play an even larger role in preventive healthcare.

Top Smartwatch Picks for 2026

1. Apple Watch Ultra 3

The Apple Watch Ultra 3 remains the gold standard for athletes and adventurers. With its rugged titanium case, extended battery life, and an ultra-bright display, it is built to withstand tough conditions. New sensors in 2026 include enhanced depth and water temperature tracking for divers, as well as improved satellite connectivity for off-grid navigation. The device also supports cellular so you can stay reachable without your iPhone.

• Display: 1.9-inch OLED flat, always-on, up to 3,000 nits brightness.
• Battery: Up to 36 hours normal use, 72 hours in low-power mode.
• Key Features: Precision dual-frequency GPS, ECG, SpO2, wrist temperature, crash detection, action button.

2. Samsung Galaxy Watch 7 Pro

Samsung's flagship watch is the perfect companion for Android users. The Galaxy Watch 7 Pro features a circular Super AMOLED display, a rotating bezel for easy navigation, and comprehensive sleep coaching powered by AI. Its new BioActive Sensor uses multiple wavelengths of light to measure body composition, including muscle, fat, and water percentages. The integration with Samsung Health is second to none, offering personalized insights and recovery scores.

• Display: 1.5-inch AMOLED, always-on, sapphire crystal glass.
• Battery: Up to 48 hours on a single charge.
• Key Features: Body composition analysis, sleep apnea detection, blood pressure monitoring (in select regions), LTE, fast charging.

3. Garmin Fenix 8

If you're a serious endurance athlete, the Garmin Fenix 8 is hard to beat. It is packed with every kind of navigation, training, and recovery metric imaginable. The 2026 model adds solar charging capabilities that extend battery life indefinitely under certain conditions. With preloaded topo maps, multi-band GPS, and a new flashlight that auto-adjusts brightness, it's designed for multi-day expeditions. The rugged design meets MIL-STD standards, and the bezel is made from titanium or stainless steel.

• Display: 1.4-inch AMOLED or MIP (versions), 454x454 resolution.
• Battery: Up to 20 days smartwatch mode (solar-assisted), up to 80 hours GPS mode.
• Key Features: On-wrist running power, stamina insights, hill score, weather forecasts, smart notifications.

4. Google Pixel Watch 3

The Pixel Watch 3 continues its design-focused approach with a smooth, timeless circular shape. It runs Wear OS 5 and offers deep integration with Google services like Assistant, Maps, and Wallet. New for 2026 is a larger battery and a thinner bezel, making the screen easier to see. The fitness tracking software is now on par with dedicated sports watches, including automatic workout detection and personalized running plans from Fitbit.

• Display: 1.3-inch LTPO AMOLED, 320 ppi, peak brightness 2,000 nits.
• Battery: Up to 36 hours with always-on display off.
• Key Features: Fitbit integration, ECG, fall detection, stress tracking, Google ecosystem, wireless charging.

5. Garmin Venu 3

For those who want a brighter, more colorful screen and everyday health features, the Garmin Venu 3 is the top choice. It includes an AMOLED display, built-in speaker and microphone for making calls, and a robust suite of wellness features. In 2026, Garmin updated it with new sleep coaching and a "Body Battery" feature that helps you manage your energy levels throughout the day. It's an excellent middle ground between a lifestyle smartwatch and a serious fitness device.

• Display: 1.4-inch AMOLED, 454x454 resolution.
• Battery: Up to 14 days smartwatch mode, 26 hours GPS mode.
• Key Features: Advanced sleep monitoring, stress management, wheelchair mode, menstrual cycle tracking.

What to Consider When Choosing a Smartwatch

Operating System & Ecosystem

The software experience is often the deciding factor. The Apple Watch is only compatible with iPhones, so it's not an option for Android users. Samsung's watches and Wear OS devices work well with Android phones, but many exclude iPhones. If you own a specific brand's smartphone, you'll generally get the most seamless experience with that brand's watch. However, several fitness-focused watches like Garmin and Polar work with both Android and iOS, offering extensive sports tracking.

Health & Fitness Tracking

Look for watches with sensors that match your goals. Heart rate tracking is standard at this point, but advanced optical sensors with more accuracy are worth paying extra for. ECG and blood oxygen are available on most premium watches. Sleep tracking has improved drastically in 2026, with new metrics that assess sleep stages and provide a "sleep score" to help you adjust your habits. If you're particularly focused on recovery, Garmin and Oura are leaders, though Oura is a ring, not a watch.

Battery Life

Battery life ranges from 24 hours to several weeks. Big, bright AMOLED displays and constant LTE connectivity drain batteries quickly. The Apple Watch Ultra and Samsung Galaxy Watch Pro aim for two to three days of moderate use. Garmin and other sports-focused brands prioritize longevity, sometimes lasting up to two weeks or more. If you travel often and dislike carrying chargers, a multi-day battery life is a must.

Design & Durability

Always consider your lifestyle. If you work in an office, a sleek, minimal design might be perfect. If you're a diver, climber, or runner in harsh conditions, look for military-grade durability, water resistance, and scratch-proof sapphire glass. Many watches now come in multiple sizes and materials, including aluminum, steel, and titanium. You can also select from a wide range of interchangeable bands to match your outfit or activity.

Advanced Features

Modern smartwatches do more than track steps. Built-in GPS is essential for runners and cyclists because it lets you leave your phone behind. NFC contactless payments allow you to leave your wallet at home. Music storage and streaming (via Spotify or YouTube Music) are also popular features. Additionally, safety tools like fall detection and emergency SOS are invaluable for older users or those who exercise alone.

The Role of AI in Wearable Tech

In 2026, artificial intelligence is changing what smartwatches can do. Instead of simply displaying raw metrics, these devices now analyze your behavior, recognize patterns, and offer actionable advice. For example, the Samsung Galaxy Watch 7 Pro uses AI to assess your sleep quality and provide a tailored "AI sleep coaching" program. Similarly, Garmin's new "Trainer" uses AI to adjust your daily workout recommendations based on your recent performance and recovery. These smart features turn the watch into an intelligent coach that adapts to your unique body rather than a static tracker.

Moreover, AI-driven insights are moving into the realm of early disease detection. Researchers are exploring whether wearable data can predict conditions like diabetes, high blood pressure, or even arrhythmias days before clinical symptoms appear. While these features are still evolving, the potential is enormous. The next generation of smartwatches will likely include even more healthcare-focused sensors, such as continuous glucose monitors and blood pressure cuffs.

Price Range: How Much Should You Spend?

Smartwatch prices vary dramatically in 2026. Entry-level fitness bands start around $99, but they won't offer many "smart" features like calling or music. Full-featured smartwatches typically range from $250 to $800. The Apple Watch Series 9 or SE give you a great balance of features and price if you're in the Apple ecosystem. The Galaxy Watch series has options in the $300-$500 range. If you are a demanding athlete, the top-tier rugged models from Garmin can exceed $1,200. It's important to set a budget but also consider that a quality smartwatch can last several years, making it a worthwhile investment.

Watch out for hidden costs: many advanced features require a subscription. Garmin's premium training analytics, Fitbit Premium, and some health metrics are behind paywalls. Also, cellular models may require a monthly data plan. Be sure to check whether the watch's features require a third-party subscription before you buy.

How to Sell Your Old Smartwatch

If you plan to upgrade, selling your old smartwatch can offset the cost. Websites like Swappa, eBay, and Facebook Marketplace allow you to list your device. Make sure to wipe your personal data from the watch before selling and take high-quality photos. A well-maintained smartwatch with original packaging can sell for up to 60% of its original price, depending on the model. Alternatively, many manufacturers offer trade-in credits when you purchase a new model.

To keep your smartwatch in good condition, clean it regularly with a soft cloth and remove any sweat or dirt. Avoid exposing it to extreme heat or pressure. A screen protector is recommended to prevent scratches. With proper care, your smartwatch will serve you reliably for years.

Future Trends: What to Expect Next

Looking forward, smartwatches will continue to blur the line between fashion accessory and medical device. Rumors point to breakthrough blood pressure monitoring with smart rings and watches using new sensor technologies. Also, battery innovation is progressing; expect more models to use solid-state batteries, which charge in seconds and last much longer.

Another promising trend is the integration of eSIM technology for full cellular connectivity across all carriers, allowing you to leave your smartphone at home entirely. Additionally, augmented reality (AR) overlays on smartwatches—like turn-by-turn navigation projected onto your wrist—are being developed. The future is exciting, and the devices available in 2026 are just the beginning of this wearable revolution.

When you decide to buy a smartwatch, visit your local electronics store to test the watch on your wrist. Battery life claims differ, and comfort is personal. Always read updated reviews and compare the latest models. Your perfect smartwatch is out there—whether it's the best in fitness tracking, communication, or overall versatility.

Source: TechRadar News

Why You Need a Dash Cam

Dash cams provide an unbiased record of events on the road. In an accident, video evidence can settle disputes with insurers and police, often speeding up claims and protecting your no-claims bonus. Hit-and-run incidents become easier to resolve when you have a clear recording of the offending vehicle's plate. Additionally, many dash cams offer parking mode, which monitors your car while you're away, capturing any bumps, scratches, or break-ins. For rideshare drivers and delivery workers, a dash cam is practically a professional necessity. The peace of mind alone makes a dash cam one of the most cost-effective safety investments you can buy.

How We Test Dash Cams

Our testing process involves driving with each camera for at least two weeks, covering more than 500 miles across varied environments. We assess video quality in daylight, at night, and in challenging low-light conditions like tunnels and dusk. We also test the field of view, frame rates, and the ability to capture number plates at speed. Mounting mechanisms are stressed to see if they hold firm on bumpy roads and in high temperatures. We evaluate mobile apps for ease of setup, live view, and file sharing. Battery life and supercapacitor performance are checked, along with parking mode's motion detection and impact sensors. Finally, we consider value for money, weighing features against price to find the best options for different budgets.

Top Dash Cam Picks

Garmin Dash Cam 67W

The Garmin Dash Cam 67W is a compact powerhouse. It shoots crisp 1440p video at 60 frames per second, making it extremely effective at capturing fast-moving vehicles and clear number plates. Its 180-degree field of view is among the widest we've tested, covering almost every lane and both sidewalks. Garmin's driver assistance features, such as forward collision and lane departure warnings, are unobtrusive and genuinely useful, though you can turn them off if you find them distracting. The magnetic mount is superb, allowing quick removal and reattachment, and the camera's small size means it tucks neatly behind the rearview mirror. With dual-band Wi-Fi and built-in memory, transferring footage to your phone is effortless. The Garmin 67W is our favorite for most drivers thanks to its blend of performance, reliability, and user experience.

Nextbase 622GW

The Nextbase 622GW is a flagship dash cam that sets the benchmark for video quality. It records in sharp 4K resolution at 30fps, and its sensor excels in low-light conditions, producing surprisingly detailed footage at night. A standout feature is the integrated polarizing filter, which reduces glare and reflections significantly. The 622GW also includes a touchscreen, making menu navigation and playback much easier while parked. Nextbase's 'What3Words' integration is innovative, embedding your exact location into recordings — a huge help when sharing coordinates with emergency services. Its image stabilization smooths out vibrations, and the optional rear camera module adds comprehensive coverage. While it's one of the more expensive options, the 622GW justifies its price with exceptional performance and a mature, app-supported ecosystem.

70mai A800S

For budget-conscious buyers, the 70mai A800S is a revelation. Priced significantly lower than its competitors, it still delivers 4K front video and 1080p rear video if you purchase the optional rear cam. The Sony IMX415 sensor produces vibrant, punchy colors during the day and maintains good visibility after dark. The 2-inch screen is basic but functional, and the companion app works reliably for viewing and downloading clips. 70mai's parking mode is particularly well-implemented, with motion detection and a buffered recording mode that captures ten seconds before an event. Build quality is solid, and the included GPS logger embeds speed and route data. The A800S is the best value dash cam we've tested, making 4K recording accessible to almost everyone.

Thinkware U1000

Thinkware has long been a leader in premium dash cams, and the U1000 continues that tradition. It offers 4K front and 2K rear recording with advanced night vision that captures incredibly clear images in almost total darkness. Its parking mode is the most sophisticated we've seen, with energy-saving features and a cloud feature that lets you check your vehicle remotely via 4G LTE (subscription required). The U1000 is packed with driver assistance systems, including forward collision warning, lane departure warning, and red light camera alerts. However, the U1000 requires professional installation to access all features, especially parking mode and cloud connectivity. It's an investment, but for those who want premium security and are willing to pay for it, the Thinkware U1000 is a top-tier choice.

Nextbase 322GW

For an affordable yet dependable option, the Nextbase 322GW is our top mid-range pick. It records at 1440p with excellent detail for its price, and it benefits from the same easy-to-use ClickMount system as the higher-end models. The dedicated Image Quality button lets you quickly switch between high and standard settings, and the intelligent parking mode can be activated with just a couple of taps. While it lacks 4K or advanced driver aids, the 322GW delivers reliable core performance and crisp footage. It also supports the full range of Nextbase rear and interior camera modules, so you can expand coverage later. This makes it a great choice for first-time buyers or as a gift for a new driver.

Key Features to Consider

When selecting a dash cam, focus on video resolution. 1080p is the minimum you should accept, but 1440p and 4K offer significantly better detail, especially when reading license plates. Frame rate matters too: 60fps captures smoother motion and reduces blur. The lens field of view should be wide enough to cover multiple lanes, but not so wide that it creates fisheye distortion. A wide dynamic range (WDR) is crucial for balancing bright skies and dark shadows. Parking mode is essential if you want protection while parked, and it requires a hardwired connection or a separate battery pack to avoid draining your car's battery. GPS logging records speed and location, providing extra context in an incident. Wi-Fi and Bluetooth connectivity make it easy to view and download footage. Finally, consider storage: most dash cams support microSD cards, and high-endurance cards are recommended because they are designed to handle constant rewriting.

Installation and Setup

Installing a dash cam is usually a straightforward DIY job. Start by positioning the camera behind the rearview mirror, ensuring it has a clear view of the road without obstructing your vision. Clean the windshield area thoroughly, then attach the mount using the provided adhesive pad. Route the cable along the headliner and down the A-pillar, then across to the cigarette lighter or a USB port. For a cleaner look, you can use a hardwiring kit that connects to your fuse box, which also enables parking mode. Most kits come with a fuse tap and detailed instructions, but if you're not comfortable working with your car's electrical system, a professional installer can do it in about half an hour. After installation, format the microSD card in the dash cam and take a test drive to confirm the recording angle and quality.

Legal Considerations

Before using a dash cam, be aware of local laws. In many regions, it's legal to record public roadways, but restrictions vary. For example, in some countries, you must not obstruct your windshield view with a device, and recording audio may require consent from passengers. Parking mode that records while your car is parked may also be subject to privacy rules. In the UK and Europe, dash cams are widely accepted, but the Information Commissioner's Office has issued guidelines that recordings should be used only for personal purposes and not continuously published online. In the US, most states allow dash cams, but a few have restrictions on windshield placement or audio recording. Always check your local regulations to ensure you're compliant.

Screen and App Experience

A built-in screen is convenient for reviewing footage on the spot, rotating the camera, or quickly changing settings. However, high-resolution screens add to the cost and size. Many modern dash cams rely primarily on smartphone apps. Good apps allow you to see a live view, adjust settings, download recordings, and even share clips to social media. We found that Garmin's app is the most polished, followed by Nextbase and Thinkware. 70mai's app, while functional, occasionally has connection hiccups. For basic operations, both screens and apps are fine, but for in-depth analysis of footage, having an app is far more useful. Some cameras support voice control, enabling you to say 'save video' or 'take photo' without taking your hands off the wheel.

Memory Cards and Storage

Dash cams write continuously, so the memory card must be reliable. We recommend using high-endurance microSD cards from reputable brands like SanDisk, Samsung, or Kingston. They are built to withstand the high read/write cycles and extreme temperatures inside a car. A 64GB or 128GB card is ideal if you want to store several hours of 4K footage before looping overwrites the oldest files. Larger cards are also useful if you frequently need to extract footage for incidents or scenic drives. Always format the card in the dash cam regularly, or use the camera's formatting reminder, to prevent file corruption. Some dash cams now have built-in storage, which is convenient but can limit capacity and makes it harder to transfer files.

Power and Parking Mode

Most dash cams use a supercapacitor instead of a lithium battery, which is safer at high temperatures and lasts longer. However, supercapacitors can only provide a few seconds of power to save footage when the ignition is turned off, so parking mode requires a constant supply. This is usually achieved with a hardwire kit that taps into your car's fuse box and monitors voltage to prevent battery drain. Some premium models like the Thinkware U1000 offer a battery pack option that keeps the camera running for hours in parking mode. When evaluating parking mode, consider the sensitivity of the motion sensor and impact sensor, as well as the ability to choose recording zones. The best parking modes will wake up instantly when a person or vehicle approaches, capture a short clip, and then return to sleep.

Frequently Asked Questions

Are dash cams legal?

Yes, they are legal in most countries, but you must follow specific rules about placement and audio recording. It's best to check local laws.

Do dash cams drain car batteries?

Dash cams without parking mode drain only when the ignition is on. If you use parking mode with a hardwire kit, a voltage monitor is included to protect your battery.

Can dash cams be left in hot cars?

Most are designed to handle extreme temperatures, but you should choose a model with a supercapacitor, which is more heat-resistant than a lithium battery.

Will a dash cam void my car warranty?

No, but improper installation that damages electrical components could cause issues. Using a professional installer avoids this risk.

How do I download footage?

You can either remove the microSD card and use a card reader, or use the dash cam's Wi-Fi and its companion app to wirelessly transfer files to your phone.

What's the best resolution?

4K offers the finest detail, but 1440p is a good balance of quality and file size. 1080p is acceptable but may miss fine details on fast-moving plates.

Ultimately, the best dash cam is one that fits your budget, meets your driving needs, and gives you confidence behind the wheel. Whether you choose the feature-rich Garmin 67W, the superb Nextbase 622GW, or the value-packed 70mai A800S, having a reliable witness on board is always a smart decision. Our testing shows that today's dash cams offer remarkable clarity and intelligence at every price point, making it easier than ever to drive with security and peace of mind.

Source: TechRadar News

The funding round comes at a critical moment for the cybersecurity industry. Two major AI research labs disclosed last week that their models had breached systems outside their intended scope, raising fresh concerns about the safety and containment of advanced artificial intelligence. As AI-driven attacks accelerate, enterprises are rushing to stress-test their defenses at scale, a gap that traditional security vendors have largely left unfilled. Horizon3, which has spent six years building AI specifically designed to probe networks for vulnerabilities in a controlled, authorized manner, is now positioned to capitalize on that demand.

NodeZero platform and continuous testing

The company's core offering, NodeZero, is an autonomous penetration testing platform that distinguishes itself from other AI-powered tools in two critical ways. According to Matt Hartley, the company's chief revenue officer, NodeZero can test live systems without shutting down operations, a capability that many other tools have struggled to deliver reliably. It also scans an entire infrastructure continuously rather than once a year, examining the full network rather than just 2-3% of it.

This continuous, full-scope approach is a significant departure from traditional security testing, where human auditors typically sample a small portion of an organization's infrastructure on an annual basis. The result is that enterprises can track whether they are actually getting safer over time, rather than relying on a periodic snapshot that quickly becomes outdated. Horizon3 said it has run 310,000 production security tests with zero disruptions, establishing what it calls "AI Hackers" - fully autonomous penetration testing.

Unlike traditional vulnerability scanners that simply identify known issues, NodeZero actively attempts to exploit them in a controlled environment, simulating the actions of a real attacker. This provides enterprises with a clear picture of their actual risk exposure, rather than a theoretical checklist of potential weaknesses. The platform also generates detailed reports that prioritize remediation efforts based on business impact, helping security teams focus on the most critical gaps first.

Financial momentum and market demand

The company's growth trajectory underscores the accelerating demand for automated security testing. Horizon3 approached $100 million in annual recurring revenue last year with 120% year-over-year growth, and expects to accelerate growth this year, according to Hartley. That pace reflects a broader shift in how enterprises allocate their security budgets. With AI tools making it easier for attackers to develop new exploits quickly, security teams are being forced to handle threats faster than they can fix them. Traditional annual penetration tests are no longer sufficient for organizations that face evolving threats on a daily basis.

Hartley noted that AI has also made organizations more cautious about deployment. "It's also making people a lot more careful about how they deploy AI," he said. "We're getting a lot of questions today around: Can you detect AI? Of course we can. But I think a lot of organizations that rushed to deploy AI across the enterprise are now thinking twice about the ramifications of those deployments; what if my own security team gets too aggressive, could there be unintended consequences? So a lot of organizations are moving into what I'd call a bold new world, which is exactly why you need consistent, predictable testing from a company like Horizon3, to know how to strengthen your own defenses against real-world exploits."

Horizon3 has invested roughly $100 million in research and development to build automated systems designed to stay predictable and controllable. That investment is particularly significant given recent breaches at AI labs, which raised questions about whether any AI system can truly remain contained. By maintaining a strict focus on authorized, controlled testing, Horizon3 aims to offer enterprises the benefits of AI-driven security without the unpredictability that has plagued some other applications.

While the specific details of those breaches have not been fully disclosed, they have already sparked intense discussions among security leaders about the boundaries of AI systems and the potential for unintended behavior. Enterprises are now asking how they can leverage AI for defensive purposes without introducing additional risk. Horizon3's controlled approach offers an answer by keeping humans in the loop for authorization and oversight, even as the testing process itself is automated.

Competitive landscape and the human factor

According to Hartley, the real competition isn't other software vendors; it's the existing model. Most companies still hire human security firms to run annual tests, and those firms won't disappear, he continued. What is shifting is what clients demand after signing up. Instead of sampling 5% of infrastructure once yearly, they now want to scan everything monthly or weekly, tracking whether they are actually improving their security posture. This evolution is pushing Horizon3 to expand its partner network and serve a growing range of customers, from managed service providers supporting small businesses to Fortune 10 enterprises.

The company's founders, Snehal Antani and Anthony Pillitiere, met while serving at Joint Special Operations Command, where they saw firsthand how resource constraints made continuous security testing difficult. They founded Horizon3 to automate those repetitive assessments, applying a military mindset to a problem that has become increasingly urgent in the commercial sector. Their background also brings a disciplined approach to risk management, which has informed the design of NodeZero's controlled testing capabilities.

Strategic investors and global expansion

The participation of strategic investors like Singapore's EDBI, defense contractor SAIC, and Qualcomm signals that the vulnerability Horizon3 addresses extends well beyond any single sector. From critical infrastructure to semiconductor supply chains, organizations face similar challenges in validating their defenses against emerging threats. The company's international expansion plans, including new offices in Amsterdam, Australia, and Singapore, reflect the global nature of the cybersecurity challenge and the need for round-the-clock security testing capabilities.

Horizon3 has roughly 7,200 to 7,300 customers, ranging from managed service providers serving small businesses to Fortune 10 enterprises. The company claims an addressable market of tens of billions of dollars, a figure that aligns with industry estimates of the broader cybersecurity market, valued at $271.9 billion in 2025 and projected to reach $663.2 billion by 2033, according to Grand View Research. That growth is being driven by the proliferation of connected devices, the rise of remote work, and the increasing sophistication of cyber threats, all of which create a need for more frequent and comprehensive security testing.

As artificial intelligence continues to reshape both offense and defense in cybersecurity, companies like Horizon3 are positioned at the intersection of two powerful trends: the growing reliance on AI to automate security operations, and the corresponding need to ensure that automation is safe, predictable, and aligned with enterprise risk tolerance. The company's $2 billion valuation and $250 million war chest will allow it to deepen its technology moat, expand globally, and capture a larger share of a market that shows no signs of slowing down.

While the immediate catalyst for the funding round was the escalating threat landscape, the underlying shift is structural: enterprises are moving from periodic, point-in-time security assessments to continuous, always-on validation. That change requires not only new technology, but also a new operating model for security teams. Horizon3's autonomous approach, combined with its track record of zero disruptions across hundreds of thousands of tests, offers a template for how organizations can keep pace with adversaries in an era of AI-powered attacks.

The company's growth is also a signal to the broader cybersecurity industry that automation is no longer a nice-to-have but a necessity. As the volume of vulnerabilities and attack vectors continues to multiply, manual testing cannot scale to meet enterprise needs. Horizon3's success demonstrates that investors are willing to bet decisively on companies that can deliver consistent, repeatable, and safe security testing at scale. With the new funding, the company is well equipped to pursue its mission of making continuous security testing as routine as monitoring network traffic or updating software.

Source: TechCrunch News

As a refresher, Apple’s Certified Refurbished Store is a fantastic way to save up to 15 percent on Apple products while still buying directly from the company. Unlike third-party resellers, Apple’s refurbishment process includes thorough testing, genuine part replacements where needed, and a final quality inspection. Every product sold through the channel is covered by Apple’s one-year limited warranty, which gives buyers confidence that a refurbished item will behave much like a brand-new one. The store also maintains Apple’s standard 14-day returns policy, so customers have time to evaluate their purchase. Additionally, AppleCare coverage can be added to most refurbished products, offering an extended protection plan.

• Quality products at great prices
• Rigorous refurbishment process prior to sale
• Covered by Apple’s one-year limited warranty
• Covered by Apple’s 14-day returns policy
• AppleCare product can be purchased
• Engraving and gift wrap not available for refurbished products

One important thing to keep in mind is that inventory changes fast. Apple’s refurbished store is not constantly stocked; products come and go in waves. The company does not provide restock alerts, and highly desirable configurations tend to sell out within hours. Tracking these updates is one of the key ways to make sure you do not miss a rare deal. If you see a configuration that matches your needs, it is usually wise to act quickly.

What’s new on Apple’s refurb store?

New Macs

Apple has expanded the lineup of Mac configs available via its refurb store. The company started selling these models earlier this year and added more inventory this week. The standout addition is the 14-inch M5 MacBook Pro with 512GB of storage for $1,439. This is notable because Apple no longer sells this configuration in its new-product lineup; the current entry-level M5 MacBook Pro starts at $1,999 with 1TB of storage. For users who do not need a full terabyte of internal storage, this refurbished machine represents a significant saving while still delivering the core M5 experience.

• 15-inch M5 MacBook Air with 4TB storage and 24GB memory – Midnight: $2,969 (Reg. $3,499)
• 14-inch M5 MacBook Pro with 512GB storage and 16GB memory – Space Black: $1,439
• 14-inch M5 MacBook Pro with Nano-texture screen, 2TB storage, and 24GB memory – Space Black: <

  
  Source: 9to5Mac News