tech
Anthropic says you can’t be mean to Claude. It’s not about feelings
Anthropic bans abusive or cruel behavior toward Claude in updated AI usage policy
Key points
- Anthropic's updated usage policy, introduced in early October, bars sustained and needless abusive or cruel behavior toward its models.
- Claude has been able to end conversations with users who are persistently harmful or abusive since last year.
- A spokesperson said the prohibition does not cover frustration, blunt feedback, or testing and research.
Why it matters. The rule could shape how users interact with chatbots, since experts quoted in the story say tone can affect AI output, though the evidence is mixed.
In early October, Anthropic added a line to its AI usage policy barring users from engaging in "sustained and needless abusive or cruel behavior" toward its models. Claude has been able to end conversations with users who are "persistently harmful or abusive" since last year, and in extreme cases users could now be permanently banned. A spokesperson told CNN that ending conversations remains the main enforcement tool, not frustration, blunt feedback or testing.
Latest updates
Anthropic says you can’t be mean to Claude. It’s not about feelings
In early October, AI giant Anthropic inserted new language into its AI usage policy. Alongside banning users from trying to get Claude, the company’s AI system, to help them create bioweapons or sell illegal drugs, the company added a new line: Users should not “engage in sustained and needless abusive or cruel behavior toward our models.” Claude has had the ability to end conversations with users who were “persistently harmful or abusive” since last year, but now some users could, in extreme cases, be permanently banned from using Anthropic’s AI systems entirely under the new policy. (Anthropic declined interview requests, but a spokesperson told CNN the main enforcement mechanism will still be Claude ending the conversations, and that the prohibition is not for people expressing frustration, blunt feedback, or testing and research.) The new policy was met with some derision. Microsoft AI CEO Mustafa Suleyman responded with a melting face emoji. Some made fun of the company for treating its AI as a conscious being with feelings. But while Anthropic, which is expected to go public via an IPO later this month, has long studied model welfare and has an entire team devoted to model “psychology,” experts say there’s more to this policy than just not wanting people to be mean to a machine. Garbage in, garbage out A longstanding foundational principle in computing and AI is that the .