AI vulnerability patching has a grading problem

T2Covered by 1 source · first reported by Help Net Security at 12 Oct, 04:45 UTC

Researchers at Northwestern University tested AI coding agents on 112 real security bugs. They found that a patch which stops a crash often fails the checks the project’s maintainers wrote for the fix. With Claude Opus 4.7, third of the patches that made the crash go away still failed those…