Using its own AI-based tools, Google says it patched more Chrome vulnerabilities last month than it had in the previous two years combined. The company reported addressing 1,072 security issues in the last two Chrome releases (June), topping the 1,036 vuln. fixed across the prior 23 browser versions over the last 24 months.
Security folks had been warning that the arrival of large language models (LLM) would turbocharge vuln. discovery — and, well, the trend seems to be unfolding. That growth pressure pushes defenders toward AI as a countermeasure; many in the sec. community now treat these systems as tools they must use, not optional extras.
A chart in Google’s technical write-up — showing a sharp upward curve in issues found — makes the point visually. For context: Chrome 145 was released in February 2025; the latest two milestones, Chrome 149 and 150, landed last month. Inside Google, each release is called a “milestone.”
Doug Turner, Chrome Development Director, told TechCrunch that LLMs have “radically changed the economics of cybersecurity, turning vulnerability discovery into an automated industrial-scale process.” He added that, by leveraging models like Gemini, “we fix vulnerabilities ahead of time, stay ahead of our adversaries, and make Chrome safer with each update.”
This isn’t unique to Google. Microsoft recently said it fixed a record 570 vuln. across its products in a single Patch Tuesday, and attributed much of the increase to its own AI tooling.
Apple’s trajectory looks different so far. Independent tallies put Apple at 482 vuln. fixed in 2026 to date; if that pace holds, the year-end total will match or beat last year and sit roughly near the count from 2015. Numbers tell one story — but they don’t reveal everything about effort, risk, or impact.
Have you changed your attitude towards AI in 2026?