AAIR: Advanced in AI Risk

An autonomous AI system must remain valid, reliable, and safe when conditions differ from its training environment. Poor generalization to unfamiliar complex situations can cause navigation errors, operational disruption, or unsafe behavior; this is a primary risk for autonomous warehouse operations. NIST identifies poor generalization beyond training conditions as an AI risk that reduces trustworthiness and emphasizes safe operation beyond knowledge limits.