AI-Assisted Researchers Chain Three Zero-Day Chrome Bugs into Full-Exploit

A team at security firm XBOW chained three then-unknown Google Chrome bugs into a single attack that runs code on a Windows computer when the user only loads a web page, the company said. Its researchers built the attack in six days by directing AI agents running OpenAI's publicly available GPT-5.6 Sol model. Google has since confirmed all three bugs, fixed them, and paid XBOW its top bounty, the company added.

The exploit demonstrates how publicly available large language models can lower the barrier for sophisticated attack development. XBOW's researchers directed AI agents to identify, analyze, and chain three distinct Chrome vulnerabilities, resulting in a working full-chain exploit that required no user interaction beyond loading a web page. Google confirmed the bugs and stated that patches have been deployed to Chrome users via automatic updates.

The findings raise concerns about the speed at which AI tools can be repurposed for offensive security work, even when the underlying vulnerabilities are previously unknown. Google's prompt confirmation and bounty payment validate the report's core claims, though specific technical details of the AI agent configuration remain outside independent verification.