Cyberspace Administration finalises new AI regulations

China’s internet regulatory body has finalised measures to promote the “healthy development” of generative AI services three months after the initial public proposal was released.

The Interim Measures, which come into effect on August 15, bring generative AI services under a broad spectrum of anti-competition, data protection, and national security laws to reflect the technology’s wide-ranging potential uses. The measures constitute one of the first major attempts by a government to regulate AI at a national scale.

The notice makes clear that generative AI service providers should not infringe on the property rights of others, addressing concerns over the mushrooming of online scams involving AI in recent months. In one widely publicised case, a Mr. Guo transferred 4.3 million RMB (558, 456 USD) to a scammer who had used deep fake technology over video call to pose as Mr. Guo’s friend. Livestreaming accounts on social media have also been caught using deep fakes of celebrities’ faces to sell more goods.

The new measures, laid out in full on the Cyberspace Administration website, apply only to institutions that provide AI services for public use. Institutions that develop and apply generative AI, including educational and cultural institutions and private enterprises, are not subject to the new rules. According to Bloomberg, the finalised version of the measures show a “friendlier approach” compared to the proposal in recognition of AI’s major role in China’s quest for global digital leadership.

Besides the threat of illegal and unethical business practices, generative AI services have huge potential to undermine China’s social control systems. Article 20 of the Interim Measures states that necessary measures will be taken to deal with any generative AI originating from outside China that does not comply with the regulations. As virtually any publicly available foreign-made AI product is likely to “endanger national security interests”, this stipulation implies ChatGPT and other rivals to China’s homegrown alternatives will remain banned for the foreseeable future (users inside China can still access banned sites and platforms via VPN).


Join our newsletter