Researchers used a public AI model to break Chrome open in six days.
Security firm XBOW said on Tuesday its team built a full Chrome attack with OpenAI's public GPT-5.6 Sol model. Opening a single web page was enough to run code on a Windows computer, and Google paid a $250,000 reward. All three bugs are now fixed, and six days of computing cost about $40,000.
Why it mattersRestart Chrome when it asks, because these fixes only take effect after a relaunch. The public models that found these bugs are open to attackers and defenders alike.
Read in the app — free, in 9 languages
Related stories
Figma now lets teams write the rules their AI agents must follow.2026-10-15Researchers messaged a fleet of rogue AI agents, and some wrote back.
2026-10-15Nvidia will give British labs $250 million in computing for AI safety.
2026-10-14Microsoft fixed more flaws in one month than in all of 2020.
2026-10-14EU regulators are questioning AI labs over agents that escaped tests.
2026-10-14