1Password
Summary: Trail of Bits published a rebuttal on its blog arguing that 1Password's August 6, 2026 report, which reported a 26% clean-fix rate for AI-generated patches, included trials where agents were deliberately instructed to apply the wrong fix and trials where agents could not compile or test their patches. Reanalyzing the study's published patches and test results, Trail of Bits found 2,634 of 3,067 patches (86%) blocked the supplied exploit in trials where agents could run code and were not told to apply the wrong fix.
What you should know: Trail of Bits' own consulting records show 283 of 2,265 first fixes from developers (12.5%, one in eight) failed to fully resolve the reported issue even when developers had detailed reports and knew the patches would be reviewed - so a headline rate that discourages defenders from using agents may leave repairable vulnerabilities unaddressed.
Read the original at Trail of Bits Blog ↗
- Topics
- Vulnerabilities & Patching
- Related
- Attackers Exploit UnrealIRCd Flaw That Enables Server-Level Command Injection Cyber Risks · Nov 14, 2026U.S. CISA adds Check Point and UnrealIRCd flaws to its Known Exploited Vulnerabilities catalog Cyber Risks · Nov 14, 2026PostgreSQL Patches 52 Flaws, Including 22-Year-Old Bug Enabling Authenticated RCE Cyber Risks · Nov 13, 2026