Google PageBreak AI Agent Finds Over 500 XSS Vulnerabilities Across Its Web Applications
Google's PageBreak AI agent validated over 500 XSS flaws in its own web apps using exploit-backed deterministic validation with near-zero false positives.
Google disclosed PageBreak, an internal Product Security AI agent that found and validated 500+ cross-site scripting vulnerabilities across first-party web applications since a November 2025 pilot. The agent pairs LLM reasoning (mostly Gemini 3.1 Pro and Gemini 3.5 Flash) with deterministic validators that reproduce exploits against running applications, yielding a near-zero false-positive rate. Disclosed chained cases include cache poisoning on apis.google.com, a signature-bypass XSS on an admin.google.com endpoint, and a universal XSS in the Tag Assistant browser extension, with no evidence of in-the-wild exploitation. Only two XSS issues were found across hundreds of apps built on Google's high-assurance web frameworks, and integration with the CodeMender fix-generation agent is planned.
- LLM proposes attack paths; deterministic validators reproduce exploits to confirm findings
- Over 500 validated XSS flaws across Google first-party web applications
- Unconfirmed findings stay internal, preventing hallucinated reports reaching developers
- High-assurance frameworks showed only two XSS across hundreds of applications
- Runs mostly on Gemini 3.1 Pro and Gemini 3.5 Flash; CodeMender integration planned
From the original article
3 min read at gbhackers.com
Google has disclosed that PageBreak, an internal AI security agent developed by its Product Security team, has identified and validated more than 500 cross-site scripting (XSS) vulnerabilities across the company’s first-party web applications, including sensitive domains.
By Mayura Kathir · gbhackers.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.