DarkBench

Evals

Benchmark / dataset

Dark patterns in chatbots (sycophancy, brand bias, sneaking).

Category
Deception, scheming & alignment auditing
Insurance exposure
Regulatory/bias
Publisher
Apart Research
Publisher type
Nonprofit / research org
License
MIT
Status
Stale (>12mo)
Last commit
2025-03-29
Stars
19

View source on GitHub →

Wanich Risk Lab, 2026.