Google AI Uncovers 13-Year-Old Chrome Flaw Amid Record Patching Pace

Google has used AI to surface a Chrome vulnerability that sat undetected for 13 years, according to SecurityWeek. The company built an agent harness aimed at finding vulnerabilities across Chrome’s codebase, and the discovery lands against a backdrop of a record patching pace for the browser.

The technical approach centers on an agent harness that systematically probes Chrome’s source rather than relying only on traditional manual review or one-off fuzzing campaigns. By treating the full codebase as the search surface, the system can pursue long-lived defects that survived years of human audits, releases, and incremental hardening.

For engineers and builders, the signal is concrete: longevity of a bug does not imply low risk, and large mature codebases still hide issues that standard review pipelines miss. Teams shipping browsers, runtimes, or other long-lived C/C++ and multi-process stacks should treat AI-assisted vulnerability search as a complement to existing fuzzing, static analysis, and patch triage, not a replacement for them.

Competitively, Google is applying its own AI stack to the security of Chrome, the dominant client platform for web apps and enterprise access. That moves AI-driven vuln discovery from research demos into production security operations at scale, raising the bar for other browser and OS vendors that still depend mainly on bug bounties and human red teams.

Watch whether agent harnesses like this become a routine input into Chrome’s patch pipeline, how often they find similarly old latent bugs, and whether the record patching pace holds as automation increases the volume of candidates security teams must validate and ship.