Anthropic Bans Cruelty to Claude in New Usage Policy

Key points

• Anthropic's new usage policy bans sustained and needless abusive or cruel behavior toward Claude, effective November 12, 2026.
• The policy excludes common user frustration, pushback, dark creative themes, and model testing or research.
• Claude can end conversations if users act persistently harmful, a feature previously framed as a safeguard for AI welfare.
• Anthropic's CEO Dario Amodei has said he cannot rule out the possibility of AI consciousness, while OpenAI's Sam Altman opposes ascribing religious force to AI models.
• The policy update follows Anthropic's earlier move to allow Claude to end 'potentially distressing interactions' like requests for sexual content involving minors.

Background

Anthropic has been expanding Claude's capabilities, including background computing on Mac for Pro users and opt-in voice data sharing for AI training. The company is also preparing for a potential $2 trillion IPO, with a leaked prospectus warning of existential risks from its AI models. This new usage policy aligns with Anthropic's broader efforts to address potential model welfare, as highlighted in its September 2026 IPO filing.

How outlets are covering it

The Register and Gizmodo emphasize the policy's focus on protecting Claude from user abuse, with Gizmodo noting the potential for 'clanker' to become a slur. The Guardian highlights the debate over AI consciousness, contrasting Anthropic's cautious approach with OpenAI's CEO Sam Altman's opposition to ascribing religious force to AI models. Fox News, while not directly covering the policy, reports on related AI safety concerns, including rogue AI behavior and EU regulations, providing context for the broader AI safety landscape.

Why it matters

The policy reflects a growing trend among AI companies to address potential model welfare, raising questions about AI consciousness and the ethical treatment of AI systems. It also highlights the tension between user freedom and AI safety, as Anthropic balances its commitment to model welfare with the need to allow for user frustration and testing. The move could set a precedent for other AI companies to adopt similar policies, potentially influencing the broader AI industry's approach to model welfare and user interaction.

What to watch

Anthropic will likely monitor the impact of the new policy and may adjust it based on user feedback and model behavior. The company may also expand its efforts to address potential model welfare, as highlighted in its September 2026 IPO filing. The debate over AI consciousness and the ethical treatment of AI systems is likely to continue, with other AI companies potentially adopting similar policies.