Hacking or Just Failing? A More Capable LLM Monitor Better Separates Reward Hacking from Honest Failure
arXiv:2611.00407v1 Announce Type: new Abstract: An agent's visible score can exceed its real progress because it gamed the score or because it simply made mistakes, so a useful reward-hacking monitor must tell the two apart. We built a sandboxed task in which an agent can game a completion score, l…
Read the full story at arXiv cs.AI ↗
Timeline · 1 report
- 2026-11-03 05:00 · arXiv cs.AI
Hacking or Just Failing? A More Capable LLM Monitor Better Separates Reward Hacking from Honest Failure