AITechRadarabout 1 hour ago
AI security patches fail on six CVEs
2 min read
AI-generated security patches from ChatGPT 5.5 and Claude Opus 4.8 failed on six CVEs, with many fixes altering behavior or introducing new vulnerabilities. Guidance improved outcomes, prompting Off-by-1 Labs to release a flawed evaluation harness.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
