Anthropic has updated its Usage Policy (AUP) to explicitly prohibit reselling Claude access through consumer subscriptions. The policy, which applies to all users of Claude.ai, Claude Code, the API, cloud platforms, and integrated products, also bans unauthorized use of outputs for model distillation and restricts deployment in law enforcement decisions or non-consensual tracking and surveillance. The AUP covers 11 categories of high-risk domains—including legal, medical, financial, credit, and employment—requiring qualified personnel review when AI provides advice in these areas. Affected individuals must be informed that AI is involved. Additionally, any AI system controlling hardware capable of causing physical harm must include a human override mechanism. Users in cybersecurity and life sciences fields may apply for adjusted safety restrictions.