SAD (Situational Awareness Dataset)

Evals

Benchmark / dataset

Measures model self-knowledge and situational awareness.

Category
Deception, scheming & alignment auditing
Insurance exposure
Model failure
Publisher
Academic (Laine et al.)
Publisher type
Academic
License
CC-BY-4.0
Status
Stale (>12mo)
Last commit
2024-12-14
Stars
56

View source on GitHub →


Wanich Risk Lab, 2026.