Anthropic has updated its usage policy for Claude AI. The company now bans sustained and needless abusive behavior toward its models. The new rule targets extreme cases of repeated cruelty. It does not cover normal frustration, criticism, or model testing.
Anthropic adds new rule for Claude
Anthropic introduced the policy change on October 8. The update addresses several risks linked to the growing use of AI systems. One of the most unusual changes focuses on how users treat Claude. Anthropic bans ‘cruel’ behavior against its Claude AI when the conduct is repeated and serves no clear purpose. The company said the rule applies to sustained and needless abusive behavior. It is designed for extreme situations rather than ordinary conversations. Users can still express frustration with Claude. They can also challenge its answers or provide critical feedback. The policy also allows dark creative themes and AI testing. Research activities are not covered by the new restriction.
Claude can end abusive conversations
Anthropic had already introduced a related feature before this policy update. The company gave Claude the ability to end certain conversations with abusive users. That feature applies only to rare and extreme interactions. Anthropic said Claude’s ability to end these conversations remains the main enforcement method. This means the new policy does not necessarily introduce a major change for most users. Instead, it formally adds abusive treatment to the company’s usage rules.
AI consciousness remains uncertain
The policy comes as researchers debate whether advanced AI could eventually become conscious. Anthropic CEO Dario Amodei has said the company does not know whether AI models are conscious. However, he has remained open to the possibility. Anthropic has previously researched possible AI welfare. The company has said it remains highly uncertain about the moral status of AI models. The latest policy does not claim that Claude is conscious. It instead takes a cautious approach to possible future concerns. This position has also drawn criticism from some technology leaders and public figures. Microsoft AI chief Mustafa Suleyman has rejected the idea that current AI systems are conscious. He has argued that AI does not feel or experience suffering.
Anthropic also updates other AI safety rules
The latest policy update covers more than user behavior toward Claude. Anthropic also clarified restrictions involving election interference, weapons development, and surveillance. The company added rules for AI systems connected to hardware capable of physical action. Such systems must have qualified human oversight. An operator must be able to monitor the equipment and stop it when necessary. The equipment must also maintain a safe state if Claude disconnects. These changes reflect Anthropic’s broader focus on managing risks from increasingly capable AI systems. For now, mainly in extreme cases. The company continues to allow normal criticism, testing, and creative use.
The policy adds another layer to the growing debate over AI safety and potential AI welfare.
Claude’s latest update brings new features and changes for users.