Anthropic's Supported Regions Policy carries an effective date of November 12, 2026. It says Claude may not be used by people physically in an unsupported region, by entities incorporated or headquartered in one, or by entities majority-owned or controlled from one, wherever their staff are. The policy does not name China. It lists the places that are supported and says any country or region not on the list is unsupported. Mainland China, Hong Kong, Macau and Russia are not on the list. Taiwan is. Status as of November 13, 2026.
Who Is Covered
| Category in the policy | What it covers | Example |
|---|---|---|
| Physical location | People physically in an unsupported region, including users and staff of entities that are otherwise supported | An employee of a United States company working from Shanghai |
| Place of incorporation or headquarters | Entities incorporated or headquartered in an unsupported region, and their users and staff, even when those people are in a supported region | A Shenzhen-headquartered company's team in London |
| Ownership or control | Entities majority-owned or controlled, directly or indirectly, by persons or entities in unsupported regions, and their users and staff, even in a supported region | A Singapore subsidiary majority-owned by a mainland parent |
The examples are ours, to illustrate the wording. The policy applies to the Claude apps (Claude.ai, Claude Code and Cowork), the API and developer platform, and Claude models offered through cloud platforms and authorised resellers.
How the Policy Got Here
| Date | What Anthropic published | Test for companies |
|---|---|---|
| Before September 2025 | Terms barred use from certain regions | Location. Anthropic later said companies from those regions were still reaching its services through subsidiaries elsewhere |
| September 4, 2025 | Announcement of updated restrictions on unsupported regions | Entities more than 50% owned, directly or indirectly, by companies headquartered in unsupported regions, wherever they operate |
| October 8, 2026 | Usage Policy update post, effective November 12, 2026. Says the Supported Regions page was updated "to clarify" how the policy is enforced | Summarised as majority-owned or controlled by persons or entities in those regions |
| November 12, 2026 | Supported Regions Policy effective date | Majority-owned or controlled, directly or indirectly, by persons or entities in unsupported regions |
What Is Different From 2025
Comparing the September 2025 announcement with the November 2026 text:
- "Or controlled" is added. The 2025 announcement described an ownership test of more than 50%. The 2026 text covers entities that are majority-owned or controlled.
- Individuals count as owners. The 2025 announcement referred to ownership by companies headquartered in unsupported regions. The 2026 text says persons or entities.
- Staff location is spelled out both ways. People in an unsupported region are covered even if their employer is supported, and people in a supported region are covered if their employer is not.
Anthropic presents this as a clarification of enforcement, not a new restriction. Its October 2026 post says the earlier ownership rule dates from "last year".
Anthropic's Stated Reason
The policy page gives no reason, and the October 2026 post gives none beyond clarifying enforcement. The reasoning on record is from the September 2025 announcement. There Anthropic said companies under the control of authoritarian governments can be legally required to share data or cooperate with intelligence services, that such companies could use its models for military or intelligence purposes or to advance their own AI development through distillation, and that the change brought its terms into line with those risks.
What the Policy Does Not Say
- "Controlled" is not defined. The text says control may be direct or indirect and gives no threshold or test beyond majority ownership.
- No enforcement method is described. One news report dated November 12, 2026 says Anthropic may ask for identity verification when an account shows signs of activity from an unsupported region, and ban accounts that fail. That is not in the policy text we read, and the report quotes no Anthropic statement.
- No count of affected accounts or companies has been published by Anthropic.
- Announcement date. Some news reports say the policy was posted on November 10. Anthropic's own post announcing the changes is dated October 8.
The Rest of the November 12 Update
The regions change took effect alongside a wider Usage Policy revision. Anthropic's post lists a consolidated section on deceptive campaigns, a narrower elections section, clarified weapons and surveillance rules, new requirements when Claude controls hardware that acts physically, and a prohibition on sustained abusive behaviour toward Claude models.
For how the same fortnight looked on the investment side of the industry, see AI funding rounds, November 2026. This page describes a published policy and is not legal advice. Organisations unsure whether they are covered should read the policy and ask Anthropic.
Methodology
The policy text and effective date are from Anthropic's Supported Regions Policy page, read on November 13, 2026. The October announcement is Anthropic's Usage Policy update post, and the 2025 rule and reasoning are from Anthropic's September 4, 2025 announcement. The identity verification claim is from TokenPost. We read the pages through a tool that summarises them, so wording here is paraphrased except for short phrases, and should be checked against the policy before being relied on.