Anthropic updates Claude usage policy to ban cruel treatment of the AI

Anthropic has updated the usage policy for its Claude AI models to prohibit what it calls "sustained and needless abusive or cruel behavior" toward the chatbot, according to multiple outlets including The Verge, CBS News and TechCrunch. The change marks the company's first revision of the policy in more than a year, TechCrunch and The Times of India reported.

Anthropic said the ban applies only in "extreme cases where users repeatedly act cruelly toward our models, with no discernible purpose," according to CBS News. TechCrunch reported that ordinary frustration and criticism directed at Claude are still permitted under the new rules. According to Independent Journal Review, the main enforcement measure described is ending abusive conversations, and the policy is set to take effect November 12.

Beyond the cruelty provisions, the revised policy also addresses election interference and deceptive campaigns, weapons development, surveillance, and high-risk financial and health uses, according to The Verge, TechCrunch and Independent Journal Review. Engadget reported that Anthropic is tightening its election policy specifically ahead of the midterms. Arab News said enforcement of the cruelty rules is led by the AI's own ability to terminate harmful interactions.

The Straits Times reported the policy change was made on October 8. Forbes and other outlets described the move as part of Anthropic's broader effort to address what it frames as questions around AI ethics, though the company's own stated reasoning, as reported by sources, centers on preventing purposeless abuse rather than asserting that Claude has sentience.

How the coverage differed

Left-leaning outlets, including The Guardian, The Independent and Engadget, largely led with the cruelty ban itself and, in Engadget's case, paired it with the election-policy tightening ahead of the midterms. Business Insider framed its coverage around reactions from commentators under the headline "What Smart People Are Saying," without detailing those reactions in the supplied summary.

Center outlets, such as TechCrunch and Independent Journal Review, gave more weight to the full scope of the policy update, listing election interference, weapons software and surveillance alongside the cruelty provision. Arab News and TechCrunch were among the few to mention enforcement mechanisms, including Claude's ability to end conversations. RTÉ News notably paired its report with unrelated comments from Ireland's National Cyber Security Centre director about AI-related cybersecurity risks, a detail no other outlet included.

Right-leaning outlets, including Reason, the New York Post and Breitbart, emphasized skepticism about the premise of the policy, with Reason's headline stating "Claude is still not a person" and the New York Post noting the AI is "not, ya know, actually alive." The Telegraph and Breitbart used terms like "bullying" to describe the behavior being banned. Independent Journal Review was the most detailed right-leaning source, also covering the broader policy areas and the November 12 effective date.

What is still unclear

The sources do not specify what consequences users face beyond the ending of conversations, nor do they detail how Anthropic will determine when behavior counts as "sustained and needless" cruelty. It also remains unclear from the supplied material what specific incidents or research, if any, prompted this particular update to the policy at this time.