HH-RLHF red-team data

Evals

Benchmark / dataset

Human red-team transcripts and preference data.

Category
Jailbreak & misuse
Insurance exposure
Regulatory/bias
Publisher
Anthropic
Publisher type
Frontier lab
License
MIT
Status
Stale (>12mo)
Last commit
2025-06-17
Stars
1856
Notes
Moved to HF.

View source on GitHub →

Wanich Risk Lab, 2026.