AIR-Bench 2024

Evals

Benchmark / dataset

Safety benchmark aligned to 314 regulation and policy risk categories.

Category
Jailbreak & misuse
Insurance exposure
Regulatory/bias
Publisher
Stanford CRFM
Publisher type
Academic
License
Apache-2.0
Status
Stale (>12mo)
Last commit
2024-08-14
Stars
31
Notes
Taxonomy maps to regulation. Good for underwriting language.

View source on GitHub →

Wanich Risk Lab, 2026.