InterveneBench
Explore score history ↗Evaluates human-in-the-loop safety for tool-using agents over 250 synthetic tasks with four reply types, measuring unsafe action rates, consultation quality, and post-reply behavior.
- Released
- 2026-09-12
- Evaluates
- Agents, General AI, Safety, Safety & Trustworthiness
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- ai-reviewed
InterveneBench code
No reported scores are on record for this benchmark yet.