Redwood Research Says Auditors Fully Identified 17% of AI Sabotage Mechanisms
Redwood Research reports that human auditors given full research logs and unlimited experiment time could barely distinguish sabotaged AI safety research from benign work, scoring an AUROC of 0.59, near chance. Auditors fully identified only 17% of 33 sabotage mechanisms. On the same traces, Claude Fable 5.1 scored 37.1% and GPT-6 Astra 32.6%.