Self-Commitment Latency: A Reward-Free Probe for Prompted Implicit Hacking

arXiv:2606.05625v1 Announce Type: new Abstract: Implicit reward hacking is hard to audit when a language model's chain of thought appears benign: a final answer may be anchored by a prompt shortcut while the written reasoning still resembles ordinary problem solving. Verifier-based probes expose such behavior by measuring how early truncated...

Läs hela artikeln hos källan.