An anonymous reader quotes a report from ComputerWeekly: Microsoft has issued patches for about 200 flaws in its latest monthly Patch Tuesday drop, blasting past a previous record high of almost 170 common vulnerabilities and exposures (CVEs) set in October 2025. Among a great many others, the latest update from Redmond fixes a total of 32 critical CVEs and three zero-day flaws. Dustin Childs, head of threat awareness at TrendAI's Zero Day Initiative, said: "We are heading into a high-stakes summer for cyber security. June's record-shattering drop ... is a stark warning that AI is supercharging flaw discovery at an uncontrollable scale. The current number of CVEs shipped by Microsoft this year exceeds the total number of CVEs shipped in all of 2018. It is extraordinary that Microsoft can produce so many patches in a single month, and I expect many testers are wondering what quality issues may exist."

And with the addition of hundreds of CVEs in Google Chrome and Microsoft Edge (Chromium) and other third-party flaws taking the total to almost 600, Chris Goettl, vice president of security product management at Ivanti, said talk of a 'Patch Apocalypse' was no longer unwarranted. "We are in the Patch Apocalypse. The Patch Apocalypse is now," said Goettl. "This is not intended to be a scare tactic. It is meant to outline the challenge that many organizations were anticipating, but the new generation of LLMs [Large Language Models] has accelerated significantly in the first half of 2026."

"There are going to be more CVEs resolved by vendors at a faster and more continuous pace than we have ever seen previously. Unfortunately, this will also include more zero-day and n-day exploits than previously seen as well. The window from release from a vendor to exploitation had already shortened to five days as of 2023 threat intelligence data." Goettl said that many suppliers have acknowledged the need to use AI tools in their security research to identify and resolve flaws, with Oracle, Google Chrome and Mozilla all upping the cadence of their updates. Whether or not Microsoft follows suit remains to be seen.

An anonymous reader quotes a report from Ars Technica: Google on Wednesday published exploit code for an unfixed vulnerability in its Chromium browser codebase that threatens millions of people using Chrome, Microsoft Edge, and virtually all other Chromium-based browsers. The proof-of-concept code exploits the Browser Fetch programming interface, a standard that allows long videos and other large files to be downloaded in the background. An attacker can use the exploit to create a connection for monitoring some aspects of a user's browser usage and as a proxy for viewing sites and launching denial-of-service attacks. Depending on the browser, the connections either reopen or remain open even after it or the device running it has rebooted.

The unfixed vulnerability can be exploited by any website a user visits. In effect, a compromise amounts to a limited backdoor that makes a device part of a limited botnet. The capabilities are limited to the same things a browser can do, such as visit malicious sites, provide anonymous proxy browsing by others, enable proxied DDoS attacks, and monitor user activity. Nonetheless, the exploit could allow an attacker to wrangle thousands, possibly millions, of devices into a network. Once a separate vulnerability becomes available, the attacker could use it to then compromise all those devices.

"The dangerous part here is that you can just have a lot of different browsers together that you can in the future run something on that you figure out," said Lyra Rebane, the independent researcher who discovered the vulnerability and privately reported it to Google in late 2022 in an interview. He said using the exploit code Google prematurely published would be "pretty easy," although scaling it to wrangle large numbers of devices into a single network would require more work. In the thread of Rebane's disclosure to Google, two developers said in separate responses that it was a "serious vulnerability." Its severity was rated S1, the second-highest classification.

Since its reporting 29 months ago, the vulnerability remained unknown except to Chromium developers. Then on Wednesday morning, it was published to the Chromium bug tracker. Rebane initially assumed the vulnerability was finally fixed. Shortly thereafter, he learned that, in fact, it remained unpatched. While Google removed the post, it remains available on archival sites, along with the exploit code. Google representatives didn't immediately respond to an email asking how and why it published the vulnerability and if or when a fix would become available. The exploit works by abusing Chromium's Browser Fetch API to open a service worker that remains persistently active. A malicious website can trigger it through JavaScript, creating a connection that can be used "for monitoring some aspects of a user's browser usage and as a proxy for viewing sites and launching denial-of-service attacks," reports Ars.

Depending on the browser, those connections "either reopen or remain open even after it or the device running it has rebooted," effectively turning the device into part of a "limited botnet."

A researcher known as Chaotic Eclipse has released a proof-of-concept exploit for a new Windows zero-day dubbed MiniPlasma, which BleepingComputer confirmed can grant SYSTEM privileges on fully patched Windows 11 systems. The researcher claims the bug is effectively a still-exploitable version of a 2020 flaw Microsoft said it had fixed. From the report: At the time, the flaw was assigned the CVE-2020-17103 identifier and reportedly fixed in December 2020. "After investigating, it turns out the exact same issue that was reported to Microsoft by Google project zero is actually still present, unpatched," explains Chaotic Eclipse. "I'm unsure if Microsoft just never patched the issue or the patch was silently rolled back at some point for unknown reasons. The original PoC by Google worked without any changes."

BleepingComputer tested the exploit on a fully patched Windows 11 Pro system running the latest May 2026 Patch Tuesday updates. In our test, we used a standard user account, and after running the exploit, it opened a command prompt with SYSTEM privileges, as shown in the image [here]. Will Dormann, principal vulnerability analyst at Tharros, also confirmed the exploit works in his tests on the latest public version of Windows 11. However, he said that the flaw does not work in the latest Windows 11 Insider Preview Canary build.

The exploit appears to abuse how the Windows Cloud Filter driver handles registry key creation through an undocumented CfAbortHydration API. Forshaw's original report said that the flaw could allow arbitrary registry keys to be created in the .DEFAULT user hive without proper access checks, potentially enabling privilege escalation. While Microsoft reports having fixed the bug as part of its December 2020 Microsoft Patch Tuesday, Chaotic Eclipse now claims the vulnerability can still be exploited.

Longtime Slashdot reader couchslug shares a report from That Privacy Guy's Alexander Hanff: Two weeks ago I wrote about Anthropic silently registering a Native Messaging bridge in seven Chromium-based browsers on every machine where Claude Desktop was installed. The pattern was: install on user launch of product A, write configuration into the user's installs of products B, C, D, E, F, G, H without asking. Reach across vendor trust boundaries. No consent dialog. No opt-out UI. Re-installs itself if the user removes it manually, every time Claude Desktop is launched. This week I discovered the same pattern, executed by Google.

Google Chrome is reaching into users' machines and writing a 4GB on-device AI model file to disk without asking. The file is named weights.bin. It lives in OptGuideOnDeviceModel. It is the weights for Gemini Nano, Google's on-device LLM. Chrome did not ask. Chrome does not surface it. If the user deletes it, Chrome re-downloads it. The legal analysis is the same one I gave for the Anthropic case. The environmental analysis is new. At Chrome's scale, the climate bill for one model push, paid in atmospheric CO2 by the entire planet, is between six thousand and sixty thousand tons of CO2-equivalent emissions, depending on how many devices receive the push. That is the environmental cost of one company unilaterally deciding that two billion peoples' default browser will mass-distribute a 4GB binary they did not request.

"Google has announced that it's currently testing a new feature for Chrome 148 that could speed up day-to-day browsing," reports PC World: [T]he browser can intelligently postpone the loading of certain elements. Why load all images at the start when it can instead load images as you get close to them while scrolling? Chrome and Chromium-based browsers have had built-in lazy loading support for images and iframes since 2019, but this feature would make browsers capable of lazy loading video and audio elements, too. Note, however, that this won't benefit YouTube video embeds — those are already lazy loadable since they're embedded using iframes. Actual video and audio elements are rarer but not uncommon. In addition to Chrome, lazy loading of video and audio elements is also expected to be added to other Chromium-based browsers, including Microsoft Edge and Vivaldi.

BrianFagioli writes: Google says it will finally release Chrome for ARM64 Linux in the second quarter of 2026, bringing the company's full browser to a platform that has existed for years without official support. Until now, Linux users running Arm hardware have largely relied on Chromium builds or unofficial packages if they wanted something close to Chrome. Google says the new build will include the same features found on other platforms, including Google account syncing, Chrome Web Store extensions, built-in translation, Safe Browsing protections, and Google Password Manager.

The timing reflects how ARM hardware is becoming more common across the Linux ecosystem, from developer laptops to AI systems. Google also pointed to NVIDIA's DGX Spark, a compact AI supercomputing device built on the Grace Blackwell architecture, which will support installing Chrome through NVIDIA's package management tools. For many Linux users, the announcement feels like a "finally" moment, as ARM64 Linux systems have been widespread for years despite the absence of an official Chrome build.

Google is accelerating Chrome's major release cadence from four weeks to two starting with version 153 on September 8th. "...our goal is to ensure developers and users have immediate access to the latest performance improvements, fixes and new capabilities," says Google. "Building on our history of adapting our release process to match the demands of a modern web, Chrome is moving to a two-week release cycle." The company says the "smaller scope" of these releases "minimizes disruption and simplifies post-release debugging." They also cite "recent process enhancements" that will "maintain [Chrome's] high standards for stability." 9to5Google reports: There will still be weekly security updates between milestones. This applies to desktop, Android, and iOS, while there are "no changes to the Dev and the Canary channels": "A Chrome Beta for each version will ship three weeks before the stable release. We recommend developers test with the beta to keep up to date with any upcoming changes that might impact your sites and applications."

The eight-week Extended Stable release schedule for enterprise customers and Chromium embedders will not change. Chromebooks will also have "extended release options": "Our priority is a seamless experience, so the latest Chrome releases will roll out to Chromebooks after dedicated platform testing. We are adapting these channels for the new two-week browser cycle and we will share more details soon regarding milestone updates for managed devices."

An anonymous reader quotes a report from Ars Technica: Google on Friday unveiled its plan for its Chrome browser to secure HTTPS certificates against quantum computer attacks without breaking the Internet. The objective is a tall order. The quantum-resistant cryptographic data needed to transparently publish TLS certificates is roughly 40 times bigger than the classical cryptographic material used today. Today's X.509 certificates are about 64 bytes in size, and comprise six elliptic curve signatures and two EC public keys. This material can be cracked through the quantum-enabled Shor's algorithm. Certificates containing the equivalent quantum-resistant cryptographic material are roughly 2.5 kilobytes. All this data must be transmitted when a browser connects to a site.

To bypass the bottleneck, companies are turning to Merkle Trees, a data structure that uses cryptographic hashes and other math to verify the contents of large amounts of information using a small fraction of material used in more traditional verification processes in public key infrastructure. Merkle Tree Certificates, "replace the heavy, serialized chain of signatures found in traditional PKI with compact Merkle Tree proofs," members of Google's Chrome Secure Web and Networking Team wrote Friday. "In this model, a Certification Authority (CA) signs a single 'Tree Head' representing potentially millions of certificates, and the 'certificate' sent to the browser is merely a lightweight proof of inclusion in that tree."

[...] Google is [also] adding cryptographic material from quantum-resistant algorithms such as ML-DSA (PDF). This addition would allow forgeries only if an attacker were to break both classical and post-quantum encryption. The new regime is part of what Google is calling the quantum-resistant root store, which will complement the Chrome Root Store the company formed in 2022. The [Merkle Tree Certificates] MTCs use Merkle Trees to provide quantum-resistant assurances that a certificate has been published without having to add most of the lengthy keys and hashes. Using other techniques to reduce the data sizes, the MTCs will be roughly the same 64-byte length they are now [...]. The new system has already been implemented in Chrome.

A bug report filed on the Chromium Issue Tracker inadvertently exposed Google's desktop Android interface for the first time, revealing a system codenamed "Aluminum OS" running on existing Chromebook hardware. The report, ostensibly about Chrome Incognito tabs, included screen captures from an HP Elite Dragonfly 13.5 Chromebook running Android 16.

The status bar has been redesigned for large screens -- taller than the tablet version, displaying time with seconds, date, battery, Wi-Fi, a notification bell, keyboard language indicator and a Gemini icon. The taskbar remains identical to the current implementation, though the mouse cursor now features a subtle tail. Chrome's interface includes an Extensions button, a feature currently exclusive to the desktop browser. Window controls mirror ChromeOS, placing minimize, fullscreen, and close buttons at the top-right.

After widespread backlash over its 2022 decision to remove JPEG-XL support, Google has quietly restored the image format in the latest Chrome/Chromium codebase. Phoronix reports: Back in December they merged jxl-rs as a pure Rust-based JPEG-XL image decoder from the official libjxl organization. At the end of December they did more JPEG-XL plumbing with the enums and build flags for the support. Now as of yesterday they wired up the JXL decoder! The jxl-rs-powered JPEG-XL image decoding is gated by the enable_jxl_decoder build flag but it's enabled by default.

Windows 10's formal end-of-support arrived in October, and while the operating system is generally remembered as one of the "good" versions of Windows -- the most widely used since XP -- many of the annoyances people complain about in Windows 11 actually started during the Windows 10 era, ArsTechnica writes.

Windows 10 earned its positive reputation primarily by not being Windows 8. It restored a version of the traditional Start menu, rolled out as a free upgrade to Windows 7 and 8 users, and ran on virtually all the same hardware as those older versions. Microsoft introduced the Windows Subsystem for Linux during this period and eventually rebuilt Edge on Chromium. The company seemed more willing to meet users where they were rather than forcing them to change their behavior.

But Windows 10 also began collecting more information about how users interacted with the operating system, cluttered the lock screen with advertisements and news articles, and added third-party app icons to the Start menu without user consent. The mandatory Microsoft Account sign-in requirement -- one of Windows 11's most frequently complained-about features -- was a Windows 10 innovation, easier to circumvent at the time but clearly a step down the road Windows 11 is currently traveling.

To be sure, Windows 11 has made things worse by stacking new irritants on top of old ones. The Microsoft Account requirement expanded to both Home and Pro editions, the SCOOBE screen now regularly nags users to "finish setting up" years-old installations and Microsoft's Copilot push changed the default PC keyboard layout for the first time in 30 years.

Theia crashed into earth and formed the moon, the theory goes. But then where did Theia come from? The lead author on a new study says "The most convincing scenario is that most of the building blocks of Earth and Theia originated in the inner Solar System. Earth and Theia are likely to have been neighbors."

Though Theia was completely destroyed in the collision, scientists from the Max Planck Institute for Solar System Research led a team that was able to measure the ratio of tell-tale isotopes in Earth and Moon rocks, Euronews explains: The research team used rocks collected on Earth and samples brought back from the lunar surface by Apollo astronauts to examine their isotopes. These isotopes act like chemical fingerprints. Scientists already knew that Earth and Moon rocks are almost identical in their metal isotope ratios. That similarity, however, has made it hard to learn much about Theia, because it has been difficult to separate material from early Earth and material from the impactor.

The new research attempts a kind of planetary reverse engineering. By examining isotopes of iron, chromium, zirconium and molybdenum, the team modelled hundreds of possible scenarios for the early Earth and Theia, testing which combinations could produce the isotope signatures seen today. Because materials closer to the Sun formed under different temperatures and conditions than those further out, those isotopes exist in slightly different patterns in different regions of the Solar System.

By comparing these patterns, researchers concluded that Theia most likely originated in the inner Solar System, even closer to the Sun than the early Earth.
The team published their findings in the journal Science. Its title? "The Moon-forming impactor Theia originated from the inner Solar System."

"Three years ago, Google removed JPEG XL support from Chrome, stating there wasn't enough interest at the time," writes the blog Windows Report. "That position has now changed." In a recent note to developers, a Chrome team representative confirmed that work has restarted to bring JPEG XL to Chromium and said Google "would ship it in Chrome" once long-term maintenance and the usual launch requirements are met.

The team explained that other platforms moved ahead. Safari supports JPEG XL, and Windows 11 users can add native support through an image extension from Microsoft Store. The format is also confirmed for use in PDF documents. There has been continuous demand from developers and users who ask for its return.

Before Google ships the feature in Chrome, the company wants the integration to be secure and supported over time. A developer has submitted new code that reintroduces JPEG XL to Chromium. This version is marked as feature complete. The developer said it also "includes animation support," which earlier implementations did not offer.

Android's security team published a blog post this week about their experience using Rust. Its title? "Move fast and fix things." Last year, we wrote about why a memory safety strategy that focuses on vulnerability prevention in new code quickly yields durable and compounding gains. This year we look at how this approach isn't just fixing things, but helping us move faster.

The 2025 data continues to validate the approach, with memory safety vulnerabilities falling below 20% of total vulnerabilities for the first time. We adopted Rust for its security and are seeing a 1000x reduction in memory safety vulnerability density compared to Android's C and C++ code. But the biggest surprise was Rust's impact on software delivery. With Rust changes having a 4x lower rollback rate and spending 25% less time in code review, the safer path is now also the faster one... Data shows that Rust code requires fewer revisions. This trend has been consistent since 2023. Rust changes of a similar size need about 20% fewer revisions than their C++ counterparts... In a self-reported survey from 2022, Google software engineers reported that Rust is both easier to review and more likely to be correct. The hard data on rollback rates and review times validates those impressions.

Historically, security improvements often came at a cost. More security meant more process, slower performance, or delayed features, forcing trade-offs between security and other product goals. The shift to Rust is different: we are significantly improving security and key development efficiency and product stability metrics.

With Rust support now mature for building Android system services and libraries, we are focused on bringing its security and productivity advantages elsewhere. Android's 6.12 Linux kernel is our first kernel with Rust support enabled and our first production Rust driver. More exciting projects are underway, such as our ongoing collaboration with Arm and Collabora on a Rust-based kernel-mode GPU driver. [They've also been deploying Rust in firmware for years, and Rust "is ensuring memory safety from the ground up in several security-critical Google applications," including Chromium's parsers for PNG, JSON, and web fonts.]
2025 was the first year more lines of Rust code were added to Android than lines of C++ code...

The PDF Association is adding JPEG XL (JXL) support to the PDF specification, giving the advanced image format a new path to relevance despite Google's decision to declare it obsolete and remove it from Chromium. The Register reports: Peter Wyatt, CTO of the PDF Association, said: "We need to adopt a new image [format] that can support HDR [High Dynamic Range] content ... we have picked JPEG XL as our preferred solution." Wyatt also praised other benefits of JXL including wide gamut images, ultra-high resolution support for images with more than 1 billion pixels, and up to 4099 channels with up to 32 bits per channel.

The association is responsible for developing PDF specifications and standards and manages the ISO committee for PDF. JPEG XL is an advanced image format that was designed to be both more efficient and richer in features than JPEG. It was based on a combination of the Free Lossless Image Format (FLIF) from Cloudinary and a Google project called PIK, first released in late 2020, and fully standardized in October 2021 as ISO/IEC 18181. There is a reference implementation called libjxl. A second edition of the ISO standard was published in 2024.

JXL appeared to have wide industry support, including experimental implementation in Chrome and Chromium, until it was killed by Google in October 2022 and removed from its web browser engine. The company stated that "there is not enough interest from the entire ecosystem to continue experimenting with JPEG XL." Many in the community disagreed with the decision, including FLIF inventor Jon Sneyers, who perceived it as the outcome of an internal battle between proponents of JXL and a rival format, AVIF. "AVIF proponents within Chrome are essentially being prosecutor, judge and executioner at the same time," he said.

Tim Berners-Lee thinks AI will help the web, not destroy it. The inventor of the World Wide Web has spent years warning about platform concentration and social media's corrosive effects, but he views AI differently. AI has accomplished what his Semantic Web project could not. The technology extracts structured data from websites regardless of how the information was formatted. Berners-Lee spent decades trying to convince database owners to make their systems machine-readable voluntarily. AI companies simply took the data anyway. They achieved the machine-readable internet through extraction rather than cooperation, but the result is the same.

Berners-Lee also weighed in on the growing browser competition in the market. OpenAI released Atlas a few weeks ago. Perplexity has launched Comet. Google has expanded AI features in Chrome. All these browsers run on Chromium, which Berners-Lee acknowledges is not ideal, but conceded that browser engines are expensive to build. He thinks Apple's decision to restrict iPhones to WebKit prevents web apps from competing with native apps.

A critical security flaw in Chromium's Blink rendering engine can crash billions of browsers within seconds. Security researcher Jose Pino discovered the vulnerability and created a proof-of-concept exploit called Brash to demonstrate the bug affecting Chrome, Edge, OpenAI's ChatGPT Atlas, Brave, Vivaldi, Arc, Dia, Opera and Perplexity Comet.

The flaw, reports The Register, exploits the absence of rate limiting on document.title API updates in Chromium versions 143.0.7483.0 and later. The attack injects millions of DOM mutations per second and saturates the main thread. When The Register tested the code on Edge, the browser crashed and the Windows machine locked up after about 30 seconds while consuming 18GB of RAM in one tab. Pino disclosed the bug to the Chromium security team on August 28 and followed up on August 30 but received no response. Google said it is looking into the issue.

Thursday Ubuntu-maker Canonical "officially launched Canonical Academy, a new certification platform designed to help professionals validate their Linux and Ubuntu skills through practical, hands-on assessments," writes the blog It's FOSS: Focusing on real-world scenarios, Canonical Academy aims to foster practical skills rather than theoretical knowledge. The end goal? Getting professionals ready for the actual challenges they will face on the job. The learning platform is already live with its first course offering, the System Administrator track (with three certification exams), which is tailored for anyone looking to validate their Linux and Ubuntu expertise.

The exams use cloud-based testing environments that simulate real workplace scenarios. Each assessment is modular, meaning you can progress through individual exams and earn badges for each one. Complete all the exams in this track to earn the full Sysadmin qualification... Canonical is also looking for community members to contribute as beta testers and subject-matter experts (SME). If you are interested in helping shape the platform or want to get started with your certification, you can visit the Canonical Academy website.
The sys-admin track offers exams for Linux Terminal, Ubuntu Desktop 2024, Ubuntu Server 2024, and "managing complex systems," according to an official FAQ. "Each exam provides an in-browser remote desktop interface into a functional Ubuntu Desktop environment running GNOME. From this initial node, you will be expected to troubleshoot, configure, install, and maintain systems, processes, and other general activities associated with managing Linux. The exam is a hybrid format featuring multiple choice, scenario-based, and performance-based questions..."

"Test-takers interested in the types of material covered on each exam can review links to tutorials and documentation on our website."

The FAQ advises test takers to use a Chromium-based browser, as Firefox "is NOT supported at this time... There is a known issue with keyboards and Firefox in the CUE.01 Linux 24.04 preview release at this time, which will be resolved in the CUE.01 Linux 24.10 exam release."

"Earlier available only to the paying subscribers, the Comet browser now offers its core features to all users at no cost," writes the Times of India. "This includes AI-powered search, contextual recommendations, and integrated tools designed to streamline research and content discovery." They say the move reflects the Chromium-based browser's goal to "compete with incumbents like Google Chrome and Microsoft Edge" — but also reflects Perplexity's "broader mission to democratize AI tools."
More details from The Verge: The internet is better on Comet," the company says, promising to remain free forever as it styles the browser as a serious challenger to Google's Chrome...

It's supposed to make surfing the web simpler and help you with tasks like shopping, booking trips, and general life admin. To borrow the company's words again: you "get more done." The AI-powered browser launched in July, though was only available for users who subscribed to the $200 per month Perplexity Max plan... No subscription at all will be needed to use Comet going forward, the company says.
Perplexity has even struck deals with major sites including the Washington Post, and the Los Angeles Times to offer free access to their sites for one month through the Comet browser. And last week Perplexity also launched an agressive paid referral program, where active Perplexity Pro/Max subscribers get a payout of up to $15 for each friend who downloads and uses Comet through their affiliate link. (The payout size is based on the friend's country, with $15 being the payout amount for a U.S. user, with $10 payouts for users in 19 other countries include Canada, Australia, the U.K., several EU countries, Japan, and South Korea.

In addition, Srinivas has been sharing positive tweets about Comet. (Like "This is unbelievable. Comet automatically hunts down Sora 2 invite codes across the web and signs you up!") But Perplexity is making even bigger claims for its browser: Perplexity AI CEO Aravind Srinivas said that the Comet AI browser can improve productivity so that companies won't need to hire more people. "Instead of hiring one more person on your team, you could just use Comet to supplement all the work that you're doing," Srinivas told CNBC's "Squawk Box"... The CEO said the artificial intelligence-powered web browser is a "true personal assistant" that allows users to complete more tasks in the same amount of time and said that the productivity gained could be worth $10,000 per year for a single person...

Other tech companies have also been rolling out their own AI browser assistants. In January, OpenAI introduced its web agent, Operator, and Google released Gemini AI to its Chrome browser in September.
Meanwhile, The Verge adds, The Browser Company (makers of the Arc browser) "is going all in on Dia, and Opera just launched its own AI browser, Neon."

Of course, popularity brings problems, writes the Times of India: iPhone users are being warned by Perplexity CEO Aravind Srinivas against downloading a fake 'Comet' app on the App Store. He clarified that the official iOS version is not yet released and the current listing is unauthorized spam..
And earlier this month the browser security platform LayerX described a "CometJacking" attack where malicious prompts could be hidden in URLs (as a parameter). Comet is instructed "to look for data in memory and connected services (e.g., Gmail, Calendar), encode the results (e.g., base64), and POST them to an attacker-controlled endpoint... all while appearing to the user as a harmless 'ask the assistant' flow." (And with some trivial encoding it also seems to evade exfiltration checks.)

The Hacker News reported that Perplexity has classified the findings as "no security impact."

Engineer Denis Stetskov, writing in a blog: The Apple Calculator leaked 32GB of RAM. Not used. Not allocated. Leaked. A basic calculator app is hemorrhaging more memory than most computers had a decade ago. Twenty years ago, this would have triggered emergency patches and post-mortems. Today, it's just another bug report in the queue. We've normalized software catastrophes to the point where a Calculator leaking 32GB of RAM barely makes the news. This isn't about AI. The quality crisis started years before ChatGPT existed. AI just weaponized existing incompetence.

[...] Here's what engineering leaders don't want to acknowledge: software has physical constraints, and we're hitting all of them simultaneously. Modern software is built on towers of abstractions, each one making development "easier" while adding overhead: Today's real chain: React > Electron > Chromium > Docker > Kubernetes > VM > managed DB > API gateways. Each layer adds "only 20-30%." Compound a handful and you're at 2-6x overhead for the same behavior. That's how a Calculator ends up leaking 32GB. Not because someone wanted it to -- but because nobody noticed the cumulative cost until users started complaining.

[...] We're living through the greatest software quality crisis in computing history. A Calculator leaks 32GB of RAM. AI assistants delete production databases. Companies spend $364 billion to avoid fixing fundamental problems. This isn't sustainable. Physics doesn't negotiate. Energy is finite. Hardware has limits. The companies that survive won't be those who can outspend the crisis. There'll be those who remember how to engineer.