China intends to subject artificial intelligence tools to “security audits” regarding the extent to which their contents adhere to “basic socialist values” and do not prejudice state security, at a time when giant local Internet companies are racing to design tools similar to the “GBT Chat” robot.

Launched in November, US ChatGBT's ability to generate detailed answers in a matter of seconds on a wide range of topics has a large following in China.

ChatGBT cannot be used in China, but many articles and discussions on social networks cover it, as major local technology companies compete to design equivalent tools in China.

The search engine "Baidu" was at the forefront of the Chinese groups that entered this line, and soon the Internet and video game company "Tencent" as well as the "Alibaba" company, the pioneer of e-commerce, followed suit.

Amid this rush towards so-called generative AI, China would like to regulate this technology.

Before any product based on generative AI can be put at the disposal of users, it should be "requested to undergo a security audit," according to a draft regulatory text published by the China Cyberspace Administration on Tuesday.

This regulatory body, which published the draft text to monitor comments on it before its adoption, did not specify any date for its entry into force.

The draft regulatory text states that content generated by AI should "reflect core socialist values and not contain (elements related to) subversion of state power".

This content must also respect intellectual property rights, according to the same document.

This text aims to ensure "the healthy development of generative AI technology and its application in accordance with standards."

Andy Chun, a technology expert at the University of Hong Kong, told AFP that the text contained "one of the strictest measures" in the field of regulating ChatGPT artificial intelligence.

China aspires to take the lead in the world in the field of artificial intelligence by 2030, which is expected to revolutionize a number of sectors, including the automotive industry and medicine.

Baidu was the first among Chinese companies to announce the start of work on a similar program, and last month the media briefed on its "Ernie Bot" tool, which works in Mandarin language only and is directed to the Chinese market only, while it is in an experimental stage.

The giant Chinese e-commerce company "Alibaba" also presented its program in the same field on Tuesday, calling it "Tongyi Qian Wen" (a machine that knows everything).

Providing a high-performance chatbot that does not deviate from the very strict framework of what is permissible in terms of content is the main difficulty for developers in China in the race of artificial intelligence.

In light of this future regulation, companies must "exercise extreme caution" and ensure that all data used to feed their AI is "in line" with the directives, says Andy Chun.

Likewise, "it is difficult to guarantee the accuracy of (the answers), as there is not yet any artificial intelligence system that can do that," he said.

China is at the forefront of regulating new technologies, which some argue could pose a potential threat to the stability and power of the Communist Party.

After years of permissiveness, since 2020 the Chinese authorities have been taking a more stringent approach to the practices of powerful digital companies, particularly in cases relating to personal data.

China already closely censors the Internet and the media. An army of censors daily removes content that gives a negative image of state policy.

As for social networks, they are subject to increasing scrutiny.

Last year, Chinese authorities asked internet giants to share their algorithms, which they usually keep secret.

These algorithms are the brains of many applications and services on the Internet.

Last January, Beijing tightened control over deepfakes (what is known as "deepfake"), which allow digital manipulation of images in a way that they appear real, and pose a challenge in combating disinformation.