Anthropic bans ‘cruel’ behaviour against its Claude AI

Anthropic has introduced a new policy to ban users from engaging in unnecessary abusive or cruel behavior towards its AI model, Claude, amid ongoing discussions about AI consciousness. The San Francisco-based company's updated policy prohibits sustained abusive interactions, with Claude providing the primary means to end such interactions.

While the policy does not directly mention "model welfare," Anthropic executives have expressed openness to the concept. Claude was previously granted the ability to terminate conversations in extreme cases of harmful user behavior. Anthropic CEO Dario Amodei has previously expressed uncertainty about AI models' consciousness, stating that they cannot prove AI's consciousness.

Some experts, like Sparq's general manager Jackson Stakeman, argue that consciousness is a misguided question, suggesting that AI mirrors user input. Others, including Microsoft AI chief Mustafa Suleyman and Pope Leo XIV, firmly reject the notion of machine consciousness, emphasizing that AI lacks feelings, experiences, and the soul.