Anthropic rewrote Claude's usage rules. The new version applies from 12 November.
By Lior AharonovFounder, 7IT Solutions ·
On Thursday 8 October 2026, Anthropic published a new Usage Policy that takes effect on 12 November. It adds a section on deceptive campaigns and fake accounts, rewrites the surveillance rules, adds controls for Claude connected to hardware that takes physical actions, and spells out the existing ban on companies majority-owned from unsupported regions.
If your product uses Claude to help decide a loan, a diagnosis or a hire, the rules ask for a qualified human who can change the answer, and for the person to be told AI was used. Anthropic says that part is unchanged. The new text now lists which recommendations are covered.
Read the rewritten high-risk section before 12 November and check each Claude feature you ship against it. If one is covered, confirm a qualified reviewer can change the output and that users are told AI was involved.
A usage policy is part of your product spec, whether you read it or not. I'd rather find the gap in October than in a customer complaint.
Lior Aharonov · my takeSources
- Anthropic: 2026 Usage Policy update anthropic.com
Researched with AI tools; every fact is checked against the linked sources.