Aegis 2.0 / Nemotron Content Safety V2

Evals

Benchmark / dataset

33k annotated human-LLM interactions for safety classification.

Category
Guardrails & safety classifiers (adjacent)
Insurance exposure
Regulatory/bias
Publisher
NVIDIA
Publisher type
Big tech
License
unknown
Status
Not on GitHub
Last commit
unknown
Notes
HF only. CC BY 4.0.

View source on GitHub →

Wanich Risk Lab, 2026.