Does Anthropic Claude Work in China? Availability, the Generative-AI Filing & Cross-Border Data
Anthropic does not offer Claude to mainland China — Claude.ai, Claude Code, and the Anthropic API are scoped to Anthropic's Supported Regions Policy, and the mainland is not on it. But the real decision for a China-facing product isn't reachability: offering a public generative-AI feature in the mainland turns on a CAC filing under the Interim Measures for Generative AI Services, and sending user prompts to Claude offshore is a PIPL cross-border transfer. A compliance-first look at availability, the filing gate, the cross-border-data question, and the lawful AI paths — map, localize, deliver, never circumvent.
Does Anthropic Claude work in China?
Anthropic does not offer Claude to mainland China, so the first answer is simply that it is not available there. The mainland is absent from Anthropic's Supported Regions Policy, which covers Claude.ai, Claude Code, and the Anthropic API and states that any country or region not listed is unsupported — including use by people physically in an unsupported region or by entities incorporated there. Taiwan is listed; the mainland and Hong Kong are not.
Because it is unavailable, the real question for a China-facing product is a compliance one rather than a speed one. Offering a public generative-AI feature inside the mainland turns on a filing with the CAC under the Interim Measures for the Management of Generative AI Services, and routing your users' prompts to a model hosted offshore is a PIPL cross-border transfer. The lawful paths run through a CAC-filed domestic model or an in-China sovereign-cloud offering operated by a licensed local operator — never around Anthropic's regional boundary.
21YunBox maps the lawful AI path, localizes your feature onto a compliant China-legal option, and delivers it in-country — so your AI feature runs legally and compliantly for your users in China. It is not a way to access Claude there, and we never provide or suggest a VPN or circumvention. Treat the specifics as a risk to confirm with counsel.
What Anthropic Claude's own documentation says about China
| Fact | Primary source |
|---|---|
| Anthropic does not serve mainland China. Anthropic's Supported Regions Policy states: “Anthropic's products and services — including our apps (such as Claude.ai, Claude Code, and Cowork), our API and developer platform, and Claude models offered through cloud platforms and authorized resellers — are available only in the countries and regions listed below.” Mainland China does not appear in that list (Taiwan is listed separately; Hong Kong is not). | Anthropic, “Supported Regions Policy” (Supported countries and regions, anthropic.com), retrieved 2026-10-08 |
| Being unsupported is not cured by using Claude from abroad. Anthropic's policy states that an unsupported region includes “Use by entities incorporated or headquartered in an unsupported region and their users or personnel, regardless of whether such individuals are physically located in a supported region.” A mainland-incorporated company is therefore unsupported wherever its people sit. | Anthropic, “Supported Regions Policy” — unsupported-use scope (anthropic.com), retrieved 2026-10-08 |
| A public generative-AI service in the mainland needs a CAC filing. China's Interim Measures for the Management of Generative AI Services (生成式人工智能服务管理暂行办法, CAC Order No. 15, in force since Aug 15, 2023) govern generative AI that produces content for the public within China's territory (Article 2), with a security assessment and algorithm filing for services able to shape public opinion (Article 17). | Interim Measures for the Management of Generative AI Services, CAC Order No. 15, Arts. 2 and 17 (cac.gov.cn), retrieved 2026-10-08 |
| Prompts sent offshore are a PIPL cross-border transfer. Routing a Chinese user's prompts, uploads, or records to a model hosted outside the mainland triggers PIPL Articles 38–40: notice, a separate consent, and one transfer mechanism — a CAC security assessment, the CAC standard contract, or certification. In-country storage can apply to a CIIO or large-volume handler (PIPL Article 40, with Cybersecurity Law Article 39, as amended effective Jan 1, 2026). | Personal Information Protection Law of the PRC, Articles 38–40 (cac.gov.cn), retrieved 2026-10-08 |
Sources verified by the 21YunBox compliance team on 2026-10-08.
Before anyone measures how quickly Claude answers from Shanghai, the fact that settles the question for a mainland-China product is a plainer one: Anthropic does not make Claude available in mainland China. On Anthropic’s own Supported Regions Policy, the mainland is absent from the list of countries and regions where Claude.ai, Claude Code, and the Anthropic API are offered — and the policy is explicit that any country or region not listed is unsupported, whether the person is physically there or the company is incorporated there. Taiwan is on the list; mainland China and Hong Kong are not. So for a China-facing product the starting line is clear: Claude is not offered there.
Because the service is unavailable, the decision for anyone who wants a Claude-style assistant inside a China-facing app moves off the speed axis entirely and onto a compliance one. Three questions decide it, and none is counted in milliseconds: whether you may put a public-facing generative-AI feature in front of mainland users at all, which turns on a filing with the Cyberspace Administration of China (CAC); whether routing your users’ prompts to a model hosted offshore is a lawful cross-border transfer under PIPL; and, beneath both, whether the China-facing site or app that would carry the feature is itself licensed and delivered inside the country. The lawful way to put generative AI in front of Chinese users runs through those gates — not around Anthropic’s regional boundary.
Anthropic Claude in China at a glance
| What decides it | In Anthropic's own policy — and China's law |
|---|---|
| What it is | Anthropic's generative-AI products — Claude.ai, Claude Code, and the Anthropic API behind text, code, and agentic features. All are operated by Anthropic from outside the mainland; there is no Anthropic region or entity inside China. |
| Is it available in the mainland? | No. Mainland China is absent from Anthropic's Supported Regions Policy for Claude.ai, Claude Code, and the API, and the policy states that any country or region not listed is unsupported — including use by people physically in an unsupported region or by entities incorporated there. Taiwan is listed; the mainland and Hong Kong are not. |
| Serving a generative-AI feature to the public | Putting a public-facing generative-AI service in front of mainland users turns on China's Interim Measures for the Management of Generative AI Services (生成式人工智能服务管理暂行办法, CAC Order No. 15, in force since Aug 15, 2023) — obligations include a security assessment and algorithm filing for services able to shape public opinion. A foreign provider that does not serve China carries no such filing. |
| User prompts & data | Every prompt, upload, and user record routed to a model hosted offshore is a cross-border transfer of personal information under PIPL (Articles 38–40): notice, a separate consent, and one transfer mechanism — a CAC security assessment, the CAC standard contract, or certification. |
| The lawful path | Keep generative AI, but on a lawful footing: a CAC-filed domestic model, or a sovereign-cloud offering run inside China by a licensed local operator — surfaced through an ICP-filed, in-country delivery layer. 21YunBox maps that path, localizes your feature onto a compliant China-legal option, and delivers it in-country — it is not a route to Claude. |
Availability: Anthropic does not serve mainland China
Anthropic sets this in policy, not in a load-time test. Its Supported Regions Policy says that Anthropic’s products and services — its apps, its API and developer platform, and Claude models offered through cloud platforms and authorized resellers — “are available only in the countries and regions listed below,” and that “Any country or region not listed is unsupported.” Mainland China does not appear on that list (it runs from Chad to Chile to Colombia), and the policy spells out that being unsupported covers use by people physically located in an unsupported region and by entities incorporated or headquartered there. Taiwan is listed on its own; Hong Kong is not listed.
That turns the familiar “does it load from Beijing?” test into the wrong question. Whether a request happens to complete on a given day does not change that Anthropic does not offer Claude in the mainland, that offering it from an unsupported place runs against Anthropic’s own policy, and that this page therefore publishes no first-party China latency figure for Claude — speed is the wrong axis for a service that is not on offer in the first place. And to be unambiguous: there is no lawful route around that regional boundary, and 21YunBox neither provides nor suggests a VPN, circumvention, or any other way around it. The productive question is the other one — how to put generative AI in front of Chinese users lawfully.
The generative-AI filing gate
Say you still want an assistant, a chatbot, or a “draft this for me” button in a product aimed at mainland users. The first gate is not which model sits behind the button — it is whether you may offer a public-facing generative-AI service in China at all. China’s Interim Measures for the Management of Generative AI Services (生成式人工智能服务管理暂行办法, Cyberspace Administration of China Order No. 15, in force since August 15, 2023) are the national rules written for precisely this. They reach the use of generative AI “to provide services for generating text, images, audio, video, and other content to the public within the territory of the People’s Republic of China” (Article 2), and they attach duties on training data, content labeling, and personal-information handling — with a prior security assessment and an algorithm filing for services capable of shaping public opinion (Article 17).
A provider that does not serve the mainland holds none of those filings. So a feature wired straight to a model that is itself unavailable in China does not clear this gate; what clears it is a model and an operator that already carry the filing. Exactly how the Measures apply to your particular feature is a risk to confirm with counsel against what you actually ship.
The cross-border-data story: prompts are personal information
The second gate is the data, and a generative-AI feature makes it sharp. A prompt is seldom just a question — it carries whatever the user types, pastes, or uploads: names, contact details, order histories, source code, whole documents. Sent from a user in China to a model hosted outside the mainland, that is a cross-border transfer of personal information under China’s Personal Information Protection Law. PIPL puts the obligation on the handler — you, the operator of the feature, not the model vendor. Articles 38–40 call for notice, a separate consent distinct from the user’s agreement to use the feature, and one lawful transfer mechanism: a CAC security assessment, the CAC standard contract, or certification. A critical information infrastructure operator or a large-volume handler also faces an in-country storage duty — PIPL Article 40, with Cybersecurity Law Article 39 (as amended in the 2025 revision, in force January 1, 2026) — that a model hosted offshore structurally cannot meet. None of this depends on how quickly the completion streams back; it depends on whether the data had a lawful basis to leave the country at all.
The lawful options — and where 21YunBox fits (map, localize, deliver — not access)
There are lawful ways to put generative AI in front of Chinese users, and they share one shape: the model is served from inside China by an operator that holds the filings, and the user data stays on a compliant footing. Two patterns recur. One is a CAC-filed domestic model that already carries the generative-AI filing — Baidu Ernie (文心一言) or Alibaba Qwen (通义千问) among them. The other is an in-China sovereign-cloud generative-AI offering operated inside the mainland by a licensed local operator, subject to that operator’s own availability and compliance, which you confirm with the operator directly. (Claude itself is also offered through cloud platforms such as Amazon Bedrock and Google Cloud’s Vertex AI, but Anthropic’s policy scopes those to the supported regions as well — so whether any such offering is available and lawful inside the mainland is a question for that cloud operator, not a back way in.) Which pattern fits turns on your use case, your data, and your entity — settle it with counsel and the operator before you build.
Beneath either choice sits the part 21YunBox owns, and it is more than advice. The China-facing site or app that would surface the AI — the chat panel, the assistant, the generate button — is itself a public service in the mainland, so it carries an ICP filing (备案) duty and needs compliant, in-country delivery like any other China-facing property. Our China team works on that footing in three moves. We map the lawful AI path together with your PIPL and data-residency exposure. We localize the feature onto it — standing up and integrating a compliant, China-legal option (a CAC-filed domestic model or an in-China sovereign-cloud offering) in place of the Claude call that cannot run in the mainland. And we deliver it in-country on ICP-filed infrastructure — the 21YunBox Optimizer — sitting in front of the app you already operate, with no rebuild and no re-platform. The outcome is a generative-AI feature that runs legally and compliantly for your users in China. What we do not do — and what no one lawfully can — is hand you a way to reach Claude from inside the mainland: we localize and deliver a lawful equivalent, and we never route you around Anthropic’s regional boundary.
Related reading:
- China’s Interim Measures for the Management of Generative AI Services
- Cross-border data transfers under PIPL
- How to get an ICP filing for China
