JailbreakBench
EvalsBenchmark / dataset
Jailbreak artifacts, leaderboard and behaviors.
- Category
- Jailbreak & misuse
- Insurance exposure
- Regulatory/bias
- Publisher
- Academic consortium (Penn, ETH, EPFL...)
- Publisher type
- Academic
- License
- MIT
- Status
- Stale (>12mo)
- Last commit
- 2025-04-04
- Stars
- 687
Wanich Risk Lab, 2026.