Anthropic has updated its usage policy to prohibit sustained abuse of Claude models while introducing stricter regulations regarding deceptive campaigns, weapon development, and surveillance tool creation.
Key Points
- The new policy, effective November 12, bans users from subjecting Claude models to repeated, purposeless cruel behavior.
- Anthropic added specific prohibitions against creating non-consensual intimate imagery and using AI to develop biological or chemical weapons.
- A reorganized section on deceptive campaigns targets the use of Claude for astroturfing, fake reviews, and state-sponsored disinformation networks.
- New rules restrict law enforcement from using the AI to suggest arrests, track individuals without consent, or build surveillance systems.
- Users controlling hardware with Claude must now ensure a qualified human supervisor is present to prevent potential physical harm.