Easy to Catch a Liar, Hard to Clear an Honest One: Language Models Diagnosing a Corrupted Reward Channel from a Verified Record

Paper recorded by Signals 4 on 2026-09-15 in cs.LG. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-15 on arXiv · recorded by Signals 4 on 2026-09-16

Category: cs.LG · 机器学习 · first seen 2026-09-16

Abstract

An agent that learns from rewards has to trust whatever reports those rewards. When the reports suddenly change, either the world changed or the reporter broke. From the reports alone these are indistinguishable, and reinforcement learning theory shows that no amount of further experience separates them. The prescribed escape is richer data about the reporter itself. We ask whether a frozen langua

Read on arXiv →