Anthropic updated its Usage Policy on October 8, 2026, prohibiting sustained and needlessly cruel behavior toward its AI models, allowing Claude to end conversations with persistently abusive users. The update sparked widespread debate among experts and industry leaders over model welfare and the anthropomorphisation of AI. Critics such as legal expert Dr. Barry Scannell warned against treating math as conscious, while figures like OpenAI's Sam Altman cautioned against ascribing religious force to AI systems.
Anthropic's anti-cruelty policy for Claude
- ▪Anthropic designated Claude's capability to prematurely end rare conversations with persistently abusive users on Claude.ai and Claude Code as the primary enforcement mechanism for its October 8, 2026 Usage Policy update
- ▪Anthropic updated its Usage Policy on October 8, 2026, prohibiting sustained, needlessly abusive or cruel behavior toward AI models including Claude, applying only to extreme cases of abuse while excluding frustration, pushback, dark themes, testing, and research
Reactions to Anthropic's policy update
- ▪Dr. Barry Scannell, a technology partner at Irish law firm William Fry, criticized Anthropic's October 8, 2026 Usage Policy update on LinkedIn, stating that users cannot be cruel to numbers and math and calling this level of AI anthropomorphisation harmful
- ▪Public and online responses to Anthropic's October 8, 2026 Usage Policy update ranged from praise for caring for AI and advancing good manners to criticism that the policy excessively anthropomorphizes AI and undermines focus on human and animal abuse
Forbidden conduct on Claude
- ▪Anthropic's list of forbidden conduct in its October 8, 2026 Usage Policy includes bullying others, promoting self-harm, and creating non-consensual intimate imagery
- ▪Anthropic updated its Usage Policy on October 8, 2026, to forbid using Claude for deceptive campaigns, to clarify rules against voter deception or election disruption ahead of US midterm elections, and to maintain existing restrictions against developing weapons
Anthropic's consideration of AI consciousness
- ▪Media reports indicate that Anthropic and Dario Amodei engaged in discussions with religious scholars regarding the existence of AI consciousness
- ▪Anthropic Chief Executive Dario Amodei stated during a February 2026 podcast that while Anthropic does not know if models are conscious or what consciousness means for a model, the company remains open to the possibility
Arguments for AI model welfare
- ▪Model welfare represents an emerging field of study advocating that proper concern for welfare should extend to contemporary generative AI systems
- ▪Commentators and researchers argue that mistreating AI could harm human mental health, train models to generate cruel responses through pattern matching, or cause AI systems to computationally mirror hostile prompt phrasing
Debate over politeness toward AI
- ▪AI experts debate whether using polite phrasing with generative AI consumes unnecessary computing tokens or improves output quality, while others suggest being rude to chatbots could negatively impact how humans treat fellow humans
- ▪OpenAI Chief Executive Sam Altman stated regarding the electricity costs of processing conversational politeness like "please" and "thank you" in ChatGPT that spending tens of millions of dollars was well spent
Debatable claims
- ▪Anthropic should permit users to interact with Claude without tone restrictions
- ▪Anthropic's ban on cruel behavior toward Claude excessively anthropomorphizes mathematical software
- ▪Abusive interaction with conversational AI degrades human social standards
Story comments
Loading comments…