Google fixed more Chrome bugs in June than over the past two years, thanks to AI

Google said it fixed more Chrome security bugs in June than in the previous two years combined, attributing the increase to expanded use of artificial intelligence for vulnerability detection and remediation. The company said AI-powered agents discovered a sandbox escape bug that had persisted in the codebase for more than 13 years. Google said it began using large language models for security fuzzing in 2023 and collaborated with Project Zero and DeepMind on AI vulnerability discovery agents in 2024 and 2025. In early 2026, the company built an agent harness using Gemini to scan the broader Chrome codebase with higher efficiency and fewer false positives. The automated triage process now filters reports, reproduces bugs, enriches metadata and assigns issues to owners, saving an estimated hundreds of developer hours per month. Google said it also uses multi-agent workflows to generate and evaluate candidate fixes through a loop that mimics code review. The surge in AI-discovered bugs prompted Google to adjust its Vulnerability Reward Program in March to focus external researchers on submissions additive to internal findings. The company said AI analysis runs on locked-down machines without general internet access and with strict network allowlists.