AI security tools uncover 1,000+ Chrome bugs, but only a few are exploited

A study of AI‑generated vulnerability reports found that more than 1,000 security flaws were identified by artificial‑intelligence tools, yet only 14 of those were actively exploited. Researchers noted that the influx of AI‑found bugs has prompted platforms such as GitHub to adjust their bug‑bounty programs, giving defenders more time to patch before attackers can act.

Separately, Google reported that its Gemini AI agents, integrated into the Chrome development workflow, closed 1,072 security issues within 60 days. The automated system scans source code, reproduces bugs, assigns severity scores and drafts patches, cutting average fix times from about 45 days to under a week and reducing manual effort by roughly 40 percent. Together, the findings show AI can dramatically accelerate vulnerability discovery and remediation, even though the rate of exploitation remains low.