Does OpenAI (ChatGPT) Work in China? Availability, the Generative-AI Filing & Cross-Border Data

For a mainland-China audience, the first thing to settle about OpenAI and ChatGPT is not how quickly a response streams — it is that OpenAI does not offer its consumer or API services to mainland China at all. China does not appear on OpenAI’s own list of supported countries and territories, and in mid-2024 OpenAI began enforcing that boundary against traffic from unsupported regions, the mainland among them. So for a China-facing product the starting point is plain: ChatGPT and the OpenAI API are not available there, and OpenAI warns that accessing or offering access from a place it does not support may result in an account being “blocked or suspended.”

Because it is not available, the real decision for anyone who wants a generative-AI feature in a China-facing app moves to a different axis entirely — a compliance one. Three gates decide it, and none is about milliseconds: whether you may offer a public-facing generative-AI service in the mainland at all, which turns on a filing with the Cyberspace Administration of China (CAC); whether sending your users’ prompts and data to a model hosted offshore is a lawful cross-border transfer under PIPL; and, underneath both, whether the China-facing site or app that would carry the feature is itself licensed and delivered in-country. The lawful way to put generative AI in front of Chinese users runs through those gates — not around OpenAI’s block.

OpenAI (ChatGPT) in China at a glance

Availability: OpenAI does not serve mainland China

OpenAI’s position is set in its own Help Center, not by a load-time test. China is not on the list of countries and territories where OpenAI supports access to ChatGPT, and the same is true of the OpenAI API — “If a location is not included in the list below, our API is not supported there.” In mid-2024 OpenAI began enforcing that boundary against traffic from unsupported regions, the mainland among them, warning that accessing or offering access from a place it does not support “may result in your account being blocked or suspended.” Taiwan appears on the list on its own; mainland China does not appear at all.

That makes the usual “does it load from Beijing?” framing beside the point. Whether a request happens to complete on a given day is not the question — OpenAI does not offer the service there, offering access from an unsupported place is against its own terms, and an account used that way can be blocked or suspended. For that reason this page publishes no first-party China latency figure for ChatGPT: speed is the wrong axis for a service that is not available in the first place. And to be unambiguous — there is no lawful route around that block, and 21YunBox neither provides nor suggests circumvention of any kind. The productive question is a different one: how to put generative AI in front of Chinese users lawfully.

The generative-AI filing gate

Suppose you do want an AI assistant, a chatbot, or a “generate” button in a product aimed at mainland users. The first gate is not which model you pick — 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 country’s first national rules for exactly this. They apply to 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 set obligations on training data, content labeling, personal-information handling, and — for services able to shape public opinion — a prior security assessment and an algorithm filing (Article 17).

A provider that does not serve mainland China holds none of those filings. So a feature wired directly to a model that is itself unavailable in the mainland does not clear this gate; the lawful route is a model and an operator that do. Whether and how the Measures apply to your specific 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. A prompt is rarely just a question — it carries whatever the user typed, uploaded, or pasted, often including names, contact details, order records, or 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 duty on the handler — you, not the model vendor: Articles 38–40 require notice, a separate consent distinct from the user’s agreement to use the feature, and one transfer mechanism — a CAC security assessment, the CAC standard contract, or certification. A critical information infrastructure operator or a large-volume handler faces an in-country storage duty under Article 40 that an offshore model cannot satisfy. None of this turns on how fast the completion streams back; it turns on whether the data had a lawful basis to leave the country at all.

The lawful options — and where 21YunBox fits (localize, deliver, advise — not access)

There are lawful ways to put generative AI in front of Chinese users, and they share a shape: the model is served from inside China by an operator that holds the filings, and the data stays on a compliant footing. Two patterns are common. One is a sovereign-cloud offering operated inside the mainland by a licensed local operator — for example, Azure OpenAI Service delivered through 21Vianet (世纪互联), subject to that operator’s own availability and compliance, which you confirm directly with them. The other is a domestic model that already holds the CAC generative-AI filing, such as Baidu Ernie (文心一言) or Alibaba Qwen (通义千问). Which fits depends on your use case, your data, and your entity — settle it with counsel and the operator before you build.

Underneath 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 window, 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 does three things on that footing: we map the lawful AI path and your PIPL and data-residency exposure; we localize your 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 call that cannot run in the mainland; and we deliver it in-country on ICP-filed infrastructure — the 21YunBox Optimizer — in front of the app you already run, with no rebuild and no re-platform. The result is an AI feature that runs legally and compliantly for your users in China. What we do not do, and what no one lawfully can, is give you a way to reach OpenAI from the mainland: we localize and deliver a lawful equivalent, we never route you around OpenAI’s block.

Get a compliance assessment →

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

RELATED TECHNOLOGIES

Anthropic Claude

Anthropic Claude

AWS Bedrock

AWS Bedrock

Azure OpenAI Service

Azure OpenAI Service

GitHub Copilot

GitHub Copilot

Google Gemini

Google Gemini

Hugging Face

Hugging Face

Microsoft Copilot

Microsoft Copilot

Midjourney

Midjourney

Mistral AI

Mistral AI

Perplexity

Perplexity