Anthropic bans sustained 'abusive or cruel behavior' toward Claude in new usage policy

Anthropic is updating its usage policy for the first time in over a year, with the changes taking effect on 12 November 2026. The most talked-about change prohibits "sustained and needless abusive or cruel behavior" toward Claude, The Verge reported on 8 October 2026.

What's changing? The model-welfare rule is tied to Anthropic's research on whether models may have an internal experience. Anthropic says it applies only to extreme cases of repeated cruelty with no discernible purpose, not to ordinary frustration, pushback, dark creative themes or testing, and that ending the conversation remains the primary enforcement mechanism. The update also adds rules on deceptive political and commercial campaigns, election interference, surveillance (including tracking people without consent), software and components that make weapons work, and a requirement that a qualified operator can observe and stop AI-controlled hardware that could cause injury.

Context: The rule drew wide reaction on X, including from Microsoft AI chief Mustafa Suleyman, who has argued that the idea of model welfare is wrong.

Sources: The Verge (Hayden Field); Anthropic