AI Breaks Chrome Security in 6 Days, Google Pays $250K Bounty

Chrome serves about 1 billion users worldwide, so its security matters a lot. Security team XBOW used AI to build a full-chain exploit for Chrome that could gain full system access in a short time.

With OpenAI's GPT-5.6 Sol model and four researchers, the team broke Chrome in just 6 days. The bugs were reported to Google and all fixed, earning a $250,000 bounty. Experts say humans would need about 6 months for similar work.

During the research, the AI searched source code, tested ideas, built proof-of-concepts, and wrote the first exploit versions, costing about $40,000 in computing power. Team members set the strategy, picked research directions, and reviewed the AI's work.

XBOW reminds users to update Chrome to the latest version to stay protected against these three fixed bugs.