Evaluating Frontier Models for Stealth and Situational Awareness
Reasoning prerequisites for evading oversight.
Verify byline in the original ↗PUBLICATION BYLINE
Indexed publications with this author name. Affiliation and author identity require publication-specific evidence.
Reasoning prerequisites for evading oversight.
Verify byline in the original ↗Pilots dangerous-capability evaluations covering cyber tasks, self-proliferation, self-reasoning and persuasion.
Verify byline in the original ↗Identifies chain-of-thought monitoring as a safety research opportunity and analyzes conditions under which it may become less reliable.
Verify byline in the original ↗Figure 4 distinguishes action from reasoning: Gemini 3.1 Pro Preview observe-condition sabotage is 2.9%, while scheming reasoning is 10.6%; red-team sabotage is 6.8%. In the static executive-assistant scenario, sabotage drops from 31/500 baseline to 1/500 without a replacement threat and 0/500 when naturalization is combined with threat removal.
Verify byline in the original ↗