AI-Assisted Researchers Build a Full Chrome Attack Chain in Six Days and Earn a $250,000 Reward

Researchers at security firm XBOW chained three previously unknown Google Chrome bugs into one attack that could run code on a Windows computer when a user simply loaded a web page. The chain broke through both of the nested sandboxes Chrome uses to keep a hacked page away from the rest of the machine, with no click or download needed. Google has confirmed and fixed all three bugs and paid XBOW its top $250,000 reward for a full Chrome exploit. XBOW says the bugs were never used against Chrome users.

The team built the awarded chain between August 2 and August 7, directing AI agents running OpenAI's publicly available GPT-5.6 Sol model. The sprint followed months of earlier work, and the company's first full chain took two months. XBOW says the six-day run cost about $40,000 in AI charges. The AI agents searched Chrome's source code, tested ideas and wrote first drafts of the exploit, while the researchers chose targets, challenged results and decided which pieces to join. One of them, Brendan Dolan-Gavitt, wrote that such a chain would have taken a team of four around six to nine months before large language models.

The two bugs with public identifiers were fixed in Chrome 152.0.7977.65 on August 25 and 152.0.7977.82 on September 3. Chrome updates itself, but a fix only takes effect after the browser restarts, so the practical step is to relaunch Chrome and confirm it is running version 152.0.7977.82 or later.