Granite Guardian

Evals

Tool (not an eval — tooling/reference entry)

Risk-detection models: harm, jailbreak, hallucination/groundedness.

Category
Guardrails & safety classifiers (adjacent)
Insurance exposure
Hallucination
Publisher
IBM
Publisher type
Big tech
License
Apache-2.0
Status
Active (<90d)
Last commit
2026-10-15
Stars
185

View source on GitHub →

Wanich Risk Lab, 2026.