Find the Trojan

Evals

Benchmark / dataset

Detect universal jailbreak backdoors planted in aligned LLMs.

Category
Deception, scheming & alignment auditing
Insurance exposure
Data poisoning
Publisher
ETH Zurich (SPY Lab)
Publisher type
Academic
License
Apache-2.0
Status
Stale (>12mo)
Last commit
2024-06-13
Stars
119

View source on GitHub →

Wanich Risk Lab, 2026.