US Plans to Limit American-made AI Models in China

May 13, 2024

38 2 minutes read

Untitled design6 — 0d4aeeec 1cec 4c45 810c e41523995a4c w1200 r1.jpg

The Biden administration plans to put limits on American-made artificial intelligence (AI) models, the Reuters news agency reports.

The aim is to restrict the technology that powers popular chatbots from countries such as China and Russia.

But China in the past year has built its own generative AI industry and has been urging its companies to avoid foreign technology.

So, how dependent is China now on U.S. AI models and how might new American policy on AI development affect China?

OpenAI and China

OpenAI’s key AI services such as ChatGPT and the DALL-E image generator, or maker, have not been officially released in mainland China. An OpenAI spokesperson told Reuters last year that it was unable to do so in some countries due to local “conditions.”

However, many companies and engineers have accessed OpenAI’s services using special tools like virtual private networks (VPNs) to hide their network addresses.

As a result, many Chinese companies have been able to build software and programs on top of OpenAI’s models. Chinese companies also often compare their own AI models against those of OpenAI.

OpenAI has shut down Chinese companies’ access to its service. Last December, OpenAI suspended the account of ByteDance, TikTok’s Chinese owner. The move came after technology website The Verge reported that ByteDance used OpenAI’s technology to develop its own AI.

Access to OpenAI’s AI models is also restricted in Hong Kong. Although OpenAI’s services are unavailable there, Microsoft, an investor and a partner of OpenAI, has released Copilot to the public.

Copilot is a generative AI service built with OpenAI’s latest technology. By partnering with Microsoft, companies in Hong Kong can also use OpenAI’s AI models.

Open source

The United States government’s plans are aimed at the export of closed source AI models. Closed source models involve software and training data that are not publicly available, sources told Reuters. Open source models would be beyond the restrictions of export controls.

However, China has been using many open source models developed in Western nations, such as Meta Platforms’ “Llama” series.

The Beijing Academy of Artificial Intelligence is a high-level research lab in China. In March, Chinese government media reported that the lab said the majority of Chinese AI models were built using Meta’s Llama models.

The lab told China’s Premier Li Qiang at the time that China “severely lacks autonomy” in the area.

01.AI is one of the most well-known AI companies in China. Last November, it faced strong criticism after the discovery that the company’s AI model Yi-34B was built on Meta’s Llama system.

That said, a large number of Chinese tech companies such as Baidu, Huawei, and iFlytek have been working to develop their own AI models.

Some of them claim that their models have become as capable as OpenAI’s latest GPT4 model in several areas.

Self-sufficiency

Chinese authorities have been discussing the need for the country to develop its own “controllable” AI technology.

State-backed newspaper China Daily suggested in a post last February on China’s Weibo site that ChatGPT could help the U.S. government spread disinformation.

China has also been quick to release rules on the use of generative AI. AI services must get government approval before being released to the public. As of January, China has approved over 40 AI models for public use but none of them were foreign AI models.

I’m John Russell.

Josh Ye reported on this story for Reuters. John Russell adapted it for VOA Learning English.

______________________________________________________