Google Says AI Helped Fix More Chrome Bugs in June Than in the Prior Two Years
*Google attributes a sharp rise in patched Chrome vulnerabilities to its use of large language models, matching a pattern already seen at Microsoft.*
Google stated that it resolved more security issues in Chrome during June than across the full two years before that. The company linked the increase directly to AI-assisted tools for bug discovery.
The claim appears in Google's security blog post. A TechCrunch report summarized the same point and noted that Microsoft has reported similar gains from LLM-based scanning. Both firms now surface and close vulnerabilities at rates that exceed prior manual processes.
No specific count of bugs was released in the public summaries. The Hacker News discussion linked back to the Google post and tracked 139 points with 161 comments, indicating developer interest in the scale of the change.
How the tools work in practice
The sources describe LLMs and related AI systems scanning code for patterns that human reviewers had missed. This produces candidate issues that engineers then verify and fix. The result is a higher volume of patches per month without a corresponding increase in headcount.
Prior to these tools, Chrome's security team operated on a steadier cadence of discoveries. The new approach compresses that timeline.
Reactions and context
The reports do not include counter-claims from other browser vendors. They simply record that Google and Microsoft both cite AI as the driver behind larger monthly fix totals.
Experts referenced in the TechCrunch piece had already flagged this trajectory two years earlier. The current data points confirm the prediction rather than contradict it.
Why it matters
Faster discovery changes the economics of browser security. Teams that adopt these tools can ship patches on shorter cycles, while slower organizations face a widening gap in exposure time. For users, the practical effect is fewer lingering flaws in the browsers they rely on daily. The Chrome case shows the shift is already measurable in one of the most widely used pieces of client software.
---
Sources:
No comments yet