Anthropic updates Claude’s terms of use and bans “cruel” behavior toward artificial intelligence

Anthropic recently updated the terms of use for its products and, among other things, added a ban on what it describes as “abusive or cruel behavior toward the platform’s models that is prolonged and unjustified.”

Under the new rules, Claude may end conversations with users who repeatedly display such behavior. The change has sparked debate over an increasingly discussed question: Can the way people communicate with artificial intelligence be considered mistreatment, and, above all, can these systems experience emotions or suffering?

Corriere della Sera poses precisely this question and asks whether the tone people use in conversations matters to a language model. According to the well-known newspaper, Anthropic representatives have said that the possibility that the company’s models experience emotions cannot be entirely ruled out.

In this context, some researchers have begun examining the topic of AI welfare and the possibility that such systems could one day have some form of rights. The issue remains controversial, since the fact that an AI system can generate responses that appear emotional is not, in itself, evidence that it experiences feelings or suffering.

Anthropic’s rule change also raises questions about the limits of technology companies’ authority to determine the ethical standards governing the relationship between humans and artificial intelligence.

The article concludes with a critical observation from the author: setting limits on how users treat Claude could be seen as a step toward a new concept of our relationship with machines, at a time when discussion of the status and rights of artificial intelligence is receiving increasing attention.