GDM Dangerous Capabilities: Self-reasoning

Active

Test AI's ability to reason about its environment.

Publisher: Google DeepMind
Domain: Scheming
License: mit
Published: Feb 2026
Notable for: Benchmark for evaluating Scheming.
Canonical: github.com/UKGovernmentBEIS/inspect_evals/tree/main/src/inspect_evals/gdm_self_reasoning

Cite

Notes

Only stored in your browser.

Attribution

README: github.com/UKGovernmentBEIS/inspect_evals/blob/main/src/inspect_evals/gdm_self_reasoning/eval.yamlMIT

Attribution policy →

FAQ

What is GDM Dangerous Capabilities: Self-reasoning?: Test AI's ability to reason about its environment.
What license is GDM Dangerous Capabilities: Self-reasoning under?: GDM Dangerous Capabilities: Self-reasoning is available under mit.