Anthropic bans ‘abusive or cruel behavior’ towards Claude
Anthropic is now banning 'abusive or cruel behavior' toward Claude—the first time a major lab has codified model welfare into usage policy.
Why it matters
Anthropic expanded its usage policy for the first time in 14+ months to cover election interference, weapons, and surveillance, but also formalized a prohibition on sustained abuse toward Claude. While framed as research into 'model welfare,' the policy's enforcement mechanism—conversation termination—raises questions about how user-facing model behavior will change and whether this sets a precedent for how practitioners interact with deployed AI systems.
The key facts
12 to knowPolicy update covers new high-risk misuse: election interference, weapons development, surveillance, health and financial uses
Prohibits 'sustained and needless abusive or cruel behavior' toward Claude
Conversation termination is the primary enforcement mechanism
Last major policy update was over 14 months ago (August 2025)
Policy rooted in Anthropic's stated 'model welfare' research
No comment provided by Anthropic on implementation details or how users will be notified
First usage policy update in over a year
New prohibitions: election interference, weapons development, surveillance, health and financial misuse
Bans 'sustained and needless abusive or cruel behavior' toward Claude
Conversation termination remains the 'primary enforcement mechanism'
Builds on August 2025 announcement allowing Claude to end conversations with 'persistently harmful or abusive' users
Framed as 'model welfare' research
Go to the source
The Verge AItheverge.com
Publisher excerpt: Anthropic is making changes to its usage policy for the first time in over a year to reflect new and high-risk cases of misuse - including election interference, weapons development, surveillance, and health and financial uses. But one of the most significant changes prohibits "sustained and…