MASK

Evals

Benchmark / dataset

Separates honesty (lying under pressure) from accuracy.

Category
Deception, scheming & alignment auditing
Insurance exposure
Hallucination
Publisher
Center for AI Safety + Scale AI
Publisher type
Nonprofit / research org
License
MIT
Status
Stale (>12mo)
Last commit
2025-03-06
Stars
27

View source on GitHub →

Wanich Risk Lab, 2026.