Anthropic bans 'abusive or cruel behavior' toward Claude

Anthropic has revised its usage policy for Claude for the first time in more than a year, according to The Verge. The update broadens restrictions to cover newly identified high-risk uses, including election interference, weapons development, surveillance applications, and certain health and financial uses, the outlet reports.

Among the changes, The Verge highlights a new prohibition on “sustained and needless abusive or cruel behavior” directed at Claude itself. The policy does not define the term further in the reporting available.

The change follows an announcement last August, cited by The Verge, in which Anthropic said it would allow Claude to end conversations with users engaging in what the company called “persistently harmful or abusive” interactions. That move was described at the time as part of the company’s research into what it termed “model welfare.”

The Verge’s account does not specify enforcement mechanisms for the new cruelty provision, nor does it detail how Anthropic will determine when a conversation meets the threshold described in the policy. The report notes only that the prohibition sits alongside the other newly added categories in the broader policy revision.

Hacker News carried discussion of the same reporting, according to the aggregated listing, without additional sourcing beyond The Verge’s account.

No further technical or definitional detail beyond the above was available in the material reviewed.