Google AI Agent Finds Over 500 XSS Vulnerabilities Across Web Apps

Google's AI security agent, PageBreak, has identified over 500 cross-site scripting (XSS) vulnerabilities across its first-party web applications. The system was launched as a pilot in November 2025 and became a formal project in January 2026. PageBreak tests suspected weaknesses against working services before product teams receive a report.

The AI agent uses dedicated validators to test JavaScript payloads and other attack paths, keeping false positives close to zero. This process involves injecting JavaScript into the relevant page and observing whether it executes. Google said this validation step is crucial in separating usable exploits from believable but faulty model-generated attack descriptions.

PageBreak also tests whether injected inputs alter database queries, expose files through path traversal, or trigger code execution. Another validator examines requests that may reach internal services. These checks let the system assess several attack classes and keep reports focused on working paths.

The system has found two XSS vulnerabilities among hundreds of applications built with high-assurance frameworks by September 4. Both cases involved internal applications or debug endpoints with missing protections. Google plans to connect PageBreak more closely with CodeMender, an agent designed to generate security fixes, to reduce the work created by a large volume of verified findings.