US tech firm OpenAI has restricted access to its artificial intelligence (AI) models through its application programming interface (API) service in China since Tuesday.

Several Chinese AI developers and users interviewed by the Global Times believe the ban will not significantly impact the domestic industry, as OpenAI’s products were not officially available in China. Instead, they see this restriction as a catalyst for Chinese developers to intensify their efforts in AI development and application.

The API Ban and Its Implications

API allows third-party developers to integrate ChatGPT’s capabilities into their applications. OpenAI announced the opening of the API in March 2023, enabling developers to incorporate ChatGPT into their services. However, OpenAI’s website now lists 188 countries and regions where its API service is officially available, excluding mainland China, Hong Kong, and Macao.

“Accessing or offering access to our services outside of the listed countries and territories may result in account suspension,” stated OpenAI on its website.

According to reports, Chinese developers could previously access OpenAI’s products through OpenAI’s official API service or Microsoft’s Azure services, the latter being the only legal access method in China. Microsoft confirmed it will continue to provide Azure OpenAI Service to eligible customers in China via models deployed outside the country.

Factors Behind the Suspension

Analysts suggest several reasons for OpenAI’s suspension. Legal compliance issues, particularly new draft rules from the US Department of the Treasury targeting investments in AI and other technology sectors in China, might be a factor. China’s stringent cybersecurity and data security laws also present complex compliance procedures and high costs for OpenAI in the Chinese market. Additionally, the need to train GPT-5 and a potential shortage of computing power might have influenced the decision.

This move could also help OpenAI combat companies that “package” its technology as their own, thus driving out firms without core technology and competitive strength.

Chinese Response and Market Resilience

Most Chinese companies are expected to adapt by transitioning to domestic large models or purchasing OpenAI’s services via Microsoft or other external providers. In the long run, this could benefit China’s independent development of large models and self-reliance.

Domestic developers like Moonshot, Zhipu AI, Baidu, Alibaba, and Zero One Wanwu have already introduced “relocation plans” for OpenAI API users. At the World AI Conference (WAIC) in Shanghai, various large model manufacturers from the Chinese mainland and Hong Kong showcased their relocation services.

Chinese developers are also preparing to “go global.” A Beijing-based fund manager noted that domestic developers are striving to catch up with the US and could set up branches in regions where OpenAI is legally available.

The True Gap Between China and the US

While both countries lead in generative AI, the US excels in cutting-edge and underlying technology, whereas China has strengths in industrial application and service design, according to Ni Kaomeng, initiator of the China AIGC Industrial Alliance. At WAIC, many Chinese enterprises demonstrated their generative-AI tools, with some products reportedly matching or exceeding the capabilities of OpenAI’s models.

However, challenges remain. Chinese products, though advanced, have yet to produce epoch-making models like OpenAI’s Sora, which can generate stable, semantically consistent long videos.

Investing in Basic Research and Talent

To close the gap with the US, Chinese experts and entrepreneurs emphasize the need for more investment in basic research and talent cultivation. iFLYTEK’s recent launch of the Starfire V4.0 model, benchmarked against GPT-4 Turbo, highlights progress but also the existing gap.

Tsinghua University’s new School of Artificial Intelligence, led by Turing Award winner Andrew Chi-Chih Yao, aims to train 100 doctoral students annually to address the shortage of top talents in AI frontier innovation research.

Balancing Competition and Cooperation

Beyond competition, experts stress the importance of US-China cooperation in AI governance. Xue Lan, dean of Schwarzman College at Tsinghua University, highlighted the need for global cooperation to address infrastructure gaps and the unprecedented challenges posed by generative AI.

“It is crucial to create a safe space for technology professionals from both countries to collaborate without fear, discussing and solving technical issues to achieve progress in AI governance and development,” Xue said.