Anthropic added a ban on prolonged and unprovoked abusive or cruel behavior toward its artificial intelligence models to its usage policy on Thursday. According to the revised policy, Claude's ability to end such interactions will remain the main way of enforcing the rule. The move comes as the philosophical debate over whether artificial intelligence systems could be conscious broadens.
The new text does not explicitly refer to the idea of model welfare: the view that these systems may be entitled to certain forms of protection usually reserved for living beings. However, Anthropic and its executives have publicly considered this possibility. The company did not immediately respond to the request for comment mentioned in the report.
The change extends an approach the company had already announced the previous year. At that time, it said it was giving Claude the ability to end conversations in rare and extreme cases of persistently harmful or abusive user behavior. The updated policy now explicitly includes the ban in the terms of use.
In February, Anthropic CEO Dario Amodei told The New York Times that he did not know whether artificial intelligence models were conscious. He added that the company remained open to that possibility. This position expresses uncertainty and does not establish that Claude feels or suffers.
Jackson Stakeman, a general manager at Sparq, an artificial intelligence services company based in Atlanta, proposed a different rationale for the policy. Speaking to AFP, he described the consciousness debate as a dead end, arguing that we cannot prove it even to one another. He suggested the mirror metaphor was more useful: in his view, the systems reflect, on a large scale, what people give them, which he considers sufficient reason for the policy change.
By contrast, Microsoft's head of artificial intelligence, Mustafa Suleyman, categorically rejected the idea of machine consciousness in an essay the previous month. He argued that artificial intelligence systems do not feel, have experiences or suffer. He also warned against granting rights and moral protections to technological entities that, in his assessment, are heading toward capabilities far beyond those of humans.
Pope Leo XIV also addressed the issue from a religious perspective in a homily on Thursday at St. Peter's Basilica in the Vatican. He contrasted the speed at which algorithms gather data with human recollection of lived experiences and meaning, which he linked to the soul. These different positions show that Anthropic's change touches both practical rules for interacting with technology and open philosophical questions about consciousness and moral protection.
Comments