Anthropic bans 'cruel' behavior against its Claude AI

Anthropic has updated its Claude usage policy to bar “sustained and needless abusive or cruel behavior” toward its AI models, aiming to limit user treatment described as needless cruelty. The policy, announced Thursday in San Francisco, says Claude’s ability to end interactions will be the primary enforcement mechanism. The update does not explicitly mention “model welfare,” the idea that AI systems could deserve protections typically reserved for living beings, but Anthropic leadership has entertained the concept. The company previously said Claude could terminate conversations in rare, extreme cases of persistently harmful interactions. The move also arrives amid wider debate over whether AI could be conscious, with Anthropic CEO Dario Amodei telling The New York Times in February he was unsure but open to the idea.