Xbow spent $40K on AI to chain Chrome sandbox escapes for $250K bounty

Xbow USA researchers demonstrated how artificial intelligence can accelerate the discovery of complex browser exploits, according to Computerworld. The team spent six days and $40,000 in AI compute costs to identify a chain of vulnerabilities in Google Chrome. This effort resulted in a $250,000 bug bounty from Google, marking the first payout under the company’s updated 2026 rules.

How did the exploit chain work?

The researchers successfully chained vulnerabilities to escape two distinct security boundaries within the browser, according to Computerworld. First, they escaped the sandbox containing the V8 JavaScript engine. Second, they broke out of Chrome’s operating-system process sandbox. This sequence allowed them to launch a Windows application directly from a malicious web page, potentially compromising the entire host system.

What role did AI play in the discovery?

The team used OpenAI’s GPT-5.6 Sol to assist in the discovery process, rather than a specialist security tool, according to Computerworld. They employed the model to search Chrome’s source code, test hypotheses, and build proofs of concept. However, the researchers maintained control over the strategy, deciding where to look and verifying the AI’s output. This demonstrates that publicly available models can aid in finding complex software flaws.

Why does this matter for operations teams?

The exploit chain could have enabled attackers to take control of approximately one billion Windows machines worldwide, according to Computerworld. This highlights the risk of browser-based attacks that bypass multiple layers of defense. As TechSec Feed reported on 2026-10-15, verifying package identities remains critical, as counterfeit apps continue to target users. This incident adds another vector for compromise that operations must monitor.

Questions to ask your suppliers

Security teams should ask vendors how they integrate artificial intelligence into their development and testing lifecycles. Determine whether suppliers use AI tools to scan their own codebases for vulnerabilities before release. Ask if their security teams manually verify AI-generated findings, as human oversight remains essential. Notify your risk management committee about these capabilities, ensuring they understand that automated tools can now rapidly identify complex exploit chains.

What to do and how to stay safe: Google Chrome

• Ensure all Chrome browsers on Windows endpoints are updated to the latest version to apply Google's fixes.
• Review web filtering policies to block access to malicious sites that could host exploit kits.
• Monitor endpoint detection logs for unexpected application launches originating from browser processes.
• Educate users to verify the authenticity of browser installations, avoiding counterfeit packages.

Step-by-step guide: Browser Security Settings for Small Business Teams

General security guidance from the TechSec Feed newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

How much did it cost to find the Chrome exploit?

Xbow researchers spent $40,000 on AI compute time to discover the vulnerability chain, according to Computerworld.

Did Google fix the vulnerabilities?

Yes, Google fixed the flaws after Xbow presented its findings, according to Computerworld.

What AI model was used?

The researchers used OpenAI’s GPT-5.6 Sol to search source code and build proofs of concept, according to Computerworld.