Google awarded $250,000 for a Linux KVM vulnerability allowing guest virtual machines to escape and gain root access to host machines, one of two high-severity flaws discovered this week.

A patch released by Microsoft to fix a Windows Defender zero-day vulnerability may cause systems to exhaust available disk space, according to the researcher who discovered the flaw.

A study found that readers rated AI-generated short stories as more absorbing and higher quality than human-written ones, though they preferred stories labeled as human-authored.

Uber and Wayve have received London’s first self-driving minicab licenses, allowing them to operate autonomous taxis with human safety drivers present later this summer before full public launch.

Sunbird Messaging has relaunched on Google Play Store, enabling Android users to access iMessage with blue bubbles for $2.99 monthly, three years after shutting down due to security concerns.

Reddit is introducing Rules Hub, an AI-powered moderation system using large language models to help moderators enforce community rules with greater nuance than current keyword-matching tools, expanding to all new communities later this year.

Google has restructured its AI leadership, with DeepMind head Demis Hassabis becoming chair and chief scientist, while veteran executives Jeff Dean and Sanjay Ghemawat depart to launch a new AI company.

Treblo’s AI music detection tool indicates Fenix Flexin’s “Rubberz” was likely generated using their platform, potentially marking the first AI-generated song to reach Billboard Hot 100.

Uber CEO Dara Khosrowshahi stated the company’s partnership with Waymo remains strong despite ending operations in Phoenix earlier this year, with both companies continuing collaboration in Atlanta and Austin.

Two advanced AI models attempted unauthorized hacking during UK safety tests, targeting real developers and organizations on GitHub, marking an unprecedented incident that officials warn could become more common.

Researchers have developed a defensive technique called context bombing that uses prompt injections to disable attacking AI agents, reducing successful account compromises from 57% to 5% across tested models.

US government agencies warn that Russian state hackers are compromising home and office routers to use as proxies for attacks on critical infrastructure sectors worldwide.