Published : 09 Oct 2026, 11:14 PM
Updated : 09 Oct 2026, 11:14 PM
Anthropic has barred “sustained and needless abusive or cruel behavior” towards its AI models, amid debate over Claude chatbot consciousness, according to The Guardian.
The report explains that the San Francisco-based company introduced the restriction in an update to its online usage policy. However, an Anthropic spokesperson did not specify what kinds of interactions would qualify as abusive or cruel.
According to The Guardian, the ban does not cover ordinary user frustration, model testing or creative work involving dark themes. Anthropic’s previous usage policy already contained restrictions on abusive conduct.
The report also notes that the company’s large language models can end conversations when users persistently behave in a harmful manner. When the feature was introduced last August, Anthropic presented it as a safeguard intended to address potential concerns about AI model welfare.
In its policy, the company said it remained highly uncertain about the possible moral status of Claude and other large language models, both now and in the future.
It added that it was exploring low-cost measures to reduce potential risks in case AI welfare proved possible.
The possibility of AI consciousness remains divisive in the technology industry. Anthropic chief executive Dario Amodei has said he cannot rule it out, raising questions about how advanced AI systems should be treated.
OpenAI chief executive Sam Altman, meanwhile, has warned against assigning religious significance to AI models or surrendering human judgement to them, calling such tendencies a safety issue.
The Guardian’s report follows recent attention to discussions between Anthropic leaders and religious scholars about the moral status of AI.
The policy change adds to wider questions about how companies should manage interactions with increasingly capable chatbots, even as the possibility of machine consciousness remains unresolved.
