Hacker News · AI·· 3 h agoAI score58
Audit finds AgentDojo scored 433 runs as no attack despite agents executing the attacker's action
Did the attack work? Auditing AI-agent security benchmark verdicts
AI summary
2 it missed 55 of 744. x replies it scored as attacked as warnings. 0 on PyPI) that audits verdicts, re-runs AgentDojo findings on a chosen model, and seals evidence for offline checksum verification; the preprint is not peer reviewed.
Source:Hacker News · AI · zenodo.org