Anthropic Usage Policy: Easy Checks Before 12 November
Home › Blog › Anthropic Usage Policy: Easy Checks Before 12 November
Last checked: 4 November 2026
The Anthropic usage policy has been updated and takes effect on 12 November 2026, according to the policy page. Anthropic makes the Claude assistant. Anyone who uses Claude, on its website or in a business tool built on it, will be covered by the new version.
What changed in the Anthropic usage policy
The policy page lists the following points that matter to non-technical users:
- A rule against sustained, needless abuse or cruelty toward the AI itself.
- Clear bans on scams, fake reviews, fake personas, impersonating real people or organisations, and false documents or credentials.
- A named list of high-risk uses, such as legal, medical, hiring, credit and housing decisions about a specific person. These need a qualified person to review the output, and the affected person must be told AI was used.
- A requirement to say at the start that customers are talking to AI when a chatbot answers them directly.
General information, internal drafting and research are listed as exceptions.
What it means for you
Everyday use for drafting, summarising and research is not the target. The rules bite when AI output decides something about a particular person, or when customers talk to an AI directly.
What to do this week
- Read the Usage Policy page before 12 November.
- List where you use AI with customers or in decisions about people, and keep a person in charge of each result. The guide on auditing your own tasks helps.
- If a chatbot answers customers, add a line saying it is AI. See approving every reply.
- Check any AI answer before you use it, using these six checks.
Sources: Anthropic, Usage Policy (effective date, prohibited uses, high-risk requirements); Anthropic’s PDF copy of the 2026 policy. This is a summary, not legal advice; read the full policy for your situation.
