Anthropic updates Usage Policy addressing abuse, deception, and high-risk uses

Anthropic published a revised Usage Policy on October 8, effective November 12, clarifying rules on deceptive activity, sustained abuse of models, autonomous physical actions, and high-risk domains including health, finance, elections, and surveillance. The changes respond to observed misuse patterns in influence operations and weapons-related queries. It tightens operational constraints on model deployment as capabilities expand.