Anthropic sets November 12 start for rule against persistent abuse of Claude
Anthropic's revised policy prohibits extreme, repeated abuse of its models while excluding ordinary frustration, criticism and research.
AnthropicAnthropic has announced a rule prohibiting persistent, purposeless abuse of its AI models. The change was published on October 8 and takes effect November 12, 2026.
The company says the restriction targets extreme interactions involving repeated cruelty without an identifiable purpose. Ordinary frustration and disagreement fall outside its scope. The exception also covers creative work with dark themes, along with testing and research.
Claude already has the ability to stop certain persistently abusive conversations in its chat service and Claude Code. The revised policy formalizes that boundary, and Anthropic says ending those exchanges will remain its main enforcement approach.
The provision is part of a broader policy revision addressing deceptive campaigns, weapons, surveillance and consequential uses of AI. Anthropic says it will continue revising the rules as its models' capabilities and associated risks evolve..
The day's biggest AI news in one short email, every morning. Plus an instant heads-up when a reset or a big release lands.