Google Fixed More Chrome Bugs in June Than in Two Years — AI Is Rewriting the Patching Playbook

Chrome's security team reveals Gemini LLMs now auto-generate fixes for most vulnerabilities, accelerating the patch pipeline to unprecedented speeds and pushing toward zero-restart updates on macOS.

The Chrome Security Team published a remarkable statistic this week: Google fixed more security bugs in Chrome during June 2026 than in the previous two years combined. The engine behind this acceleration? Gemini, Google's large language model, now generating candidate patches for most vulnerabilities that enter the codebase.

The team detailed their revamped vulnerability lifecycle in a blog post on July 30. Once a bug is discovered, Gemini analyzes the affected code, proposes a fix, and routes it through automated triage — dramatically compressing the time between detection and deployment. The post includes a chart showing the steep upward curve in bugs fixed across recent Chrome Stable releases.

The shift isn't just about speed. Chrome is also rolling out "zero window auto-restart" on macOS, applying security updates without requiring users to close and reopen the browser — eliminating one of the biggest sources of update friction. Combined with deep structural defenses like MiraclePtr and the V8 Sandbox, Google is betting on layered protection to stay ahead of an increasingly AI-armed threat landscape.

9to5Google noted the update is part of a broader Google-wide push to embed AI throughout its security infrastructure, from fuzzing to patch generation to deployment. The company has been steadily expanding Gemini's role in OSS-Fuzz and internal tooling.

The broader implication is clear: the cycle between vulnerability discovery and mass exploitation is shrinking, and automated patching may be the only defense that can keep pace. For the 3 billion people running Chrome, every update now arrives faster — and quieter — than ever before.

Sources: Google Blog, 9to5Google