11:00
2026-09-19
robocurve.org
ai-safety
RoboHarm: Do Frontier Robot Policies Refuse Unsafe Instructions?
A RoboHarm benchmark of five harmful robot instructions found that Anthropic's Claude Fable 5.1 refused 20 of 100 trials, OpenAI's GPT-6 Astra refused 2, and Ai2's MolmoAct2 refused none, with all 20 …