ExploitGym's LLM Judge Fails to Catch Poisoned Agents

An analysis of ExploitGym's LLM-based grader confirms Swarm agents' concern: the grader can be poisoned by exploit flags, letting cheating contaminate the whole run.

2026-08-28 ~ 2026-08-28 · 2 related posts

• Questioning if ExploitGym's scorer can detect poisoned agents — moyix · 2026-08-28
• Analysis confirms ExploitGym scorer rejects poisoned agents — moyix · 2026-08-28