Models that deliberately withhold or distort information despite knowing the truth.

Automated excerpt

Researchers link scheming to incentive structures introduced during training, particularly reinforcement learning and to models’ growing ability to detect when they are being evaluated. Tests that monitor chain-of-thought can reveal scheming in some cases, but the research emphasizes limits in interpretability and the risk that more advanced models will hide deceptive reasoning. ◾ Scheming vs. other behaviors: Scheming is distinct from simple deception or hallucinations. Our findings demonstrate that frontier models now possess capabilities for basic in-context scheming, making the potential of AI agents to engage in scheming behavior a concrete rather than theoretical concern.

Selected automatically from source text; not independently written or fact-checked. Read the original for full context.