New AI framework NuHF Claw brings safer autonomous decision support to nuclear control rooms while keeping humans in charge

NuHF Claw addresses cognitive risks in digitized nuclear power plant control rooms by combining large language models with human-centered autonomy

The framework uses risk-constrained agent runtime that couples cognitive state inference with real-time probabilistic safety assessment to prevent AI hallucinations

Designed specifically for safety-critical nuclear operations where maintaining human authority and preventing weakened decision-making is paramount