Google Uses AI Agents to Find and Fix 1,072 Chrome Security Vulnerabilities

Google is expanding the use of AI agents (Gemini and other models) across Chrome's security lifecycle to find and fix code weaknesses, automate triage, generate and validate patches, and accelerate security updates. The company reports that AI-assisted scans uncovered a long-standing Chrome sandbox escape vulnerability and that multi-agent workflows (including critic agents) and integrated tools have prevented numerous issues from reaching production, while also improving test-writing, CI integration, and patch frequency.