Google Deploys Gemini to Patch Over 1,000 Chrome Security Flaws in Two Months
Google has shared details on a large-scale experiment using AI to harden Chrome's codebase against security flaws. According to the company, its Gemini models were applied to review code across the browser's codebase, identifying and helping developers fix 1,072 security bugs within a 60-day window. The initiative is framed as a practical response to the scale of Chrome's user base—roughly 3.5 billion active users—where even a small number of unpatched vulnerabilities can pose significant risk.
Traditional code audits are time-intensive, often requiring specialized security engineers to manually trace potential weaknesses across millions of lines of code. By contrast, AI-driven analysis can continuously scan for patterns associated with known vulnerability classes, flagging areas that warrant closer human review. Google says this approach is not meant to replace security researchers but to complement them, prioritizing fixes and reducing the time that flaws remain unaddressed.
The company noted that the bugs addressed during this period spanned a range of severity levels, though it did not break down how many fell into high or critical categories. The effort is part of broader moves across the industry to incorporate large language models and other AI tools into software development and maintenance pipelines, particularly in security-sensitive contexts where speed and coverage matter.