arXiv cs.LGPaperClaude Watch
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception
The threat model has shifted from detecting output-level dishonesty to detecting hidden goals and unverbalized intent via internal activations. This paper's probes work at frontier model scale and beat text-only monitoring. For safety teams: this is validation that mechanistic monitoring is feasible and more reliable than behavioral baselines. The hard part remains building reliable interventions once you've detected deception.