Anyone who repeatedly abuses AI tools may soon find Claude gives them the silent treatment.
A new usage policy change from Anthropic notes how it will now crack down on “sustained and needless abusive or cruel behavior" toward its models. “The policy update is meant to apply only in extreme cases," says Anthropic, "where users repeatedly act cruelly toward our models, with no discernible purpose.”
You’d need to be particularly cruel with how you treat the models to get called out like this. It's unclear why Anthropic chose to include these rules now, and whether it's motivated by any particular behavior observed with its AI tools.
Anthropic says Claude will now end conversations with persistently abusive users rather than banning them outright. It has yet to detail whether the most extreme users will be subject to a ban, or whether everyone will be able to return to normal usage in a new conversation.
The usage policy doesn’t take effect until Nov. 12, but Claude's ability to stop conversations is now active.
This also won’t affect you acting frustrated or speaking honestly to Claude’s models. Anthropic says, “It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research."
Other policy changes include various refinements to its wording around other misuses of Claude AI. This includes a new section to help define what it calls "deceptive activity" from both commercial and political perspectives, as well as greater clarity around restrictions on using Claude to develop weapons software.
Outside the abusive behavior rules, Anthropic's other changes focus on wording and clarity for users rather than introducing new policies.