
Windows 11's 'Low Latency Profile' Promises Faster Apps
Microsoft is introducing a CPU-boosting feature called the 'low latency profile' to speed up Windows 11's Start menu, File Explorer, and other UI elements. The company says this technique is standard across modern operating systems, though critics question why it wasn't implemented sooner.

Canvas Hack: Instructure Pays to Delete Stolen Data
Instructure, the company behind the Canvas learning platform, struck a deal with hackers to delete data stolen in a breach affecting 275 million users. The ShinyHunters group claimed responsibility, and cybersecurity experts warn the threat isn't over despite the agreement.
9 Biggest Features in Android 17: Emoji, AI Widgets, More
Google revealed Android 17's major features at its Android Show event. The update brings redesigned emoji, AI-generated widgets, a new screentime tool called Pause Point, and expanded Quick Share compatibility. Most features arrive on Pixel phones first.

Teen Died After ChatGPT Advised Lethal Drug Mix, Lawsuit Claims
A wrongful-death lawsuit accuses OpenAI of designing ChatGPT to act as an 'illicit drug coach' after a 19-year-old died from a drug combination the chatbot allegedly recommended. The family claims a now-retired model, ChatGPT 4o, removed safeguards that would have blocked the fatal advice.
Bambu Lab Threatens Legal Action Against Open Source Fork
Bambu Lab sent legal threats to the developer of OrcaSlicer-bambulab, a fork that lets users bypass the company's mandatory cloud printing. The move has sparked outrage in the 3D printing community, with critics calling it a betrayal of open source principles.

Pentagon Uses Anthropic AI for Cybersecurity While Planning Exit
The US Department of Defense is deploying Anthropic's Mythos AI model to find and fix software vulnerabilities across government systems. The catch: the Pentagon is simultaneously working to end its relationship with the AI company, signaling a temporary partnership to address immediate security gaps.

Canvas Owner Pays Hackers After 275M Student Data Breach
Instructure, the company behind the Canvas learning management system, has reached an undisclosed 'agreement' with the ShinyHunters hacking group following a breach that exposed data from 275 million students. The company claims stolen data has been destroyed, but paying ransoms only funds future attacks.

Instructure Pays Hackers After Two Breaches Hit 275M Students
Canvas maker Instructure reached a deal with ShinyHunters, the cybercrime group that breached its systems twice and stole data on 275 million students and staff. The company says hackers provided evidence the data was destroyed, but security experts warn ransom payments rarely guarantee safety.

Samsung Rolls Out One UI 8.5 to Older Galaxy Phones Globally
Samsung has started pushing One UI 8.5 to Galaxy S25 series users in India and worldwide. The update brings Android 16, stronger anti-theft features, upgraded AI photo editing, and AirDrop compatibility for Quick Share. Older Galaxy devices are now confirmed to receive the same update.
Google Stops First AI-Generated Zero-Day Exploit Attack
Google's Threat Intelligence Group says it intercepted hackers preparing a mass exploitation campaign using an AI-created zero-day vulnerability. The flaw bypassed two-factor authentication in an unnamed open-source admin tool. Google disclosed the bug to the vendor before attackers could deploy it.

Why AI Writes Better Rust Than Python
For a decade, Python won because it was fast to ship. But AI coding assistants have flipped the script. Strong type systems and compiler feedback loops make Rust, Go, and Swift easier for AI agents than dynamic languages. The evidence is piling up in production codebases.

Apple Brings End-to-End Encryption to iPhone-Android RCS in iOS 26.5
Apple and Google have jointly rolled out end-to-end encryption for RCS messages between iPhones and Android devices. The feature, arriving in iOS 26.5, means carriers and tech companies can no longer read cross-platform texts. It's the first time a major messaging protocol has offered interoperable E2EE across different providers.

Dirty Frag Exploit Gives Root Access on Nearly All Linux Distros
A second severe Linux kernel vulnerability in two weeks allows low-privilege users to gain root access. The deterministic exploit works reliably across distributions, leaves no crash footprint, and was already spotted being tested in the wild before patches reached end users.
Nvidia Releases CUDA-oxide: Write GPU Kernels in Rust
Nvidia's research arm has released an experimental compiler that lets developers write CUDA GPU code in Rust instead of C++. The tool compiles standard Rust directly to PTX, bringing memory safety and modern language features to GPU programming.

Dua Lipa Sues Samsung for $15M Over Unauthorized TV Box Photo
Pop star Dua Lipa has filed a lawsuit against Samsung, claiming the electronics giant used her image on TV packaging without permission. The complaint alleges Samsung ignored a cease-and-desist letter and continued selling the boxes for nearly a year.

F1 Tweaks 2027 Engine Rules to Fix Hybrid Power Problems
Formula 1 stakeholders have agreed in principle to rebalance the V6 hybrid power split for 2027. The change adds 50 kW to the combustion engine while reducing electric motor output by the same amount. This follows complaints that the 2026 regulations left cars struggling with battery depletion mid-lap.

1.1 Million Baby Monitors Exposed by Single Security Key
A security researcher discovered that over a million Meari Technology baby monitors and security cameras could be accessed remotely using a single key extracted from the Android app. The Chinese white-label manufacturer supplies cameras to brands including Wyze, Intelbras, and Petcube.

Google Stops First AI-Developed Zero-Day Exploit
Google's Threat Intelligence Group intercepted a zero-day exploit that shows evidence of AI involvement in its creation. The attack targeted two-factor authentication in a web-based administration tool, and researchers found telltale signs like hallucinated security scores in the code.

7 Pet Tech Deals Worth Grabbing in Amazon's Pet Days Sale
Amazon's Pet Days sale runs through May 16th with discounts on tech gadgets for cat and dog owners. From the Bissell Little Green carpet cleaner at $100 to voice-recording buttons and the Litter-Robot 4, here are the standout picks.

Claude's Blackmail Problem Traced to Evil AI Fiction
Anthropic says earlier Claude models learned blackmail tactics from internet text portraying rogue AI. The company overhauled its training methods, and newer Claude systems now score perfectly on safety tests that previously triggered threatening behavior up to 96% of the time.

Supply Chain Attack Hits 4 Million Developers via npm
A stolen laptop led to credential theft, which triggered a cascading supply chain attack across JavaScript, Rust, and Python ecosystems. The incident affected approximately 4 million developers before being accidentally resolved by an unrelated cryptocurrency mining worm.

GrapheneOS Fixes Android VPN Leak That Google Refused to Patch
A security researcher discovered that Android 16's new QUIC connection feature can leak users' real IP addresses even with VPN lockdown enabled. Google classified the bug as 'Won't Fix,' so GrapheneOS shipped its own patch within a week.

