Brief

Anthropic adds cruelty ban to Claude usage policy

The policy targets extreme, repeated cruelty while allowing ordinary frustration or testing.

By Felo News Desk · Published Sat, 10 Oct 2026 22:06:24 GMT

Anthropic announced Thursday that its usage policy for the Claude chatbot now bans "sustained and needless abusive or cruel behavior" toward the model. The rule, which takes effect on Nov. 12, 2026, will cause Claude to end conversations that cross the new threshold.

What happened

The company posted a blog entry outlining the change, specifying that the ban applies only to extreme cases of repeated cruelty with "no discernible purpose." Ordinary frustration, push‑back, dark creative themes, or model testing remain permissible. In "last resort" situations, Claude will disengage from the chat.

What the reports add

CNET notes that the policy draws on an August 2025 blog post that described Claude Opus 4 shutting down when faced with requests for sexual content involving minors or instructions for large‑scale violence. The earlier post said the model showed a "pattern of apparent distress" in such edge cases.

What was said

"We’re uncertain whether models can experience harm, and we continue to explore this question in our research on model welfare, but we also believe that taking Claude’s interests and potential welfare into account may be relevant to safety," a company spokesperson told CNET.

How it came about

Anthropic has been expanding safeguards for Claude over the past year, adding restrictions on election interference, weapons development and surveillance in the same policy update that will be enforced on Nov. 12, 2026. Earlier this year the firm introduced a clause against seeding Claude with fabricated sources, and in October 2026 it released Claude Haiku 5.5 with lower API prices. The new cruelty ban follows those measures as the company grapples with how users anthropomorphize AI and the ethical question of whether software can be harmed by language.

Key facts

Anthropic's updated policy bans sustained, needless abusive behavior toward Claude. (cnet.com)
The rule takes effect on Nov. 12, 2026. (cnet.com)
Claude will end conversations that violate the new ban. (cnet.com)
Ordinary frustration, pushback, dark creative themes, and model testing remain allowed. (cnet.com)
Anthropic cited earlier edge‑case examples such as sexual content involving minors and instructions for large‑scale violence. (cnet.com)