Anthropic said on October 8 that it added a new Usage Policy restriction covering sustained, purposeless abuse toward Claude, with the update set to take effect on November 12, 2026.
In the Anthropic usage policy update, the company says the rule is meant for extreme cases where users are repeatedly cruel with no clear purpose, not ordinary frustration, adversarial prompting, dark creative work, or safety testing and research.
Anthropic also says Claude on Claude.ai and Claude Code can end rare conversations with persistently abusive users, and that ending chats remains the primary enforcement mechanism. The change sits inside a broader 2026 policy refresh that covers Anthropic products and services, including developer and API access, cloud-provider routes, resellers, and integrated products.
One point remains unclear from the public materials: whether this specific change creates a new path to account-level penalties beyond chat termination. The Verge noted the same ambiguity.