Content produced by generative artificial intelligence services should embody “core socialist values,” China’s cyberspace regulator has said as part of its latest draft measures to regulate AI technology firms.

blue bright lights
Photo: Pixabay/

In a list of proposed regulations unveiled on Tuesday, the Cyberspace Administration of China said the country supported the independent innovation of artificial intelligence. But developers of generative AI products must comply with the proposed legal requirements and “respect social morality and public order.”

AI-generated content should not contain messages that would overturn the socialist system, incite separatism, undermine national unity or promote terrorism and extremism, the authorities said. Messages that spread ethnic hatred and discrimination, or contain violent, obscene, pornographic and false information would also be prohibited.

“Content generated by generative artificial intelligence should embody core socialist values, and shall not contain content that subverts the state power,” the regulator said.

The draft rules, which are set to take effect later this year, were published on the same day as Chinese tech giant Alibaba Group announced the roll out of its new AI model Tongyi Qianwen to rival ChatGPT. The product, which has Chinese and English language capabilities, will be incorporated into all of Alibaba’s business applications in the near future, the firm said.

Alibaba artificial intelligence
Chairman and CEO of Alibaba Group and CEO of Alibaba Cloud Intelligence, Daniel Zhang, announces a new AI model to support enterprises’ intelligence transformation on April 11, 2023. Photo: Alibaba Group.

Cyberspace authorities in China suggested mandating generative AI service providers to take measures to prevent discrimination and the generation of false information. They must also respect and protect intellectual property rights, business ethnics and personal privacy.

If the generated content was found or reported to have breached the regulations, service providers should respond by filtering such messages. They should also optimise their model within three months to ensure the content in question did not appear again, the proposed rules read.

If a user was found to have violated the regulations, such as using generative AI products for “malicious posting and commenting,” sending junk mails and improper marketing, providers shall suspend or terminate the services.

Those who fail to comply with the regulations shall face criminal liability and punishment under existing laws on cybersecurity, data security and protection of personal information, the cyberspace regulator said.

Members of the public may submit their feedback on the proposed regulations via email or letter until May 10.

Support HKFP  |  Policies & Ethics  |  Error/typo?  |  Contact Us  |  Newsletter  | Transparency & Annual Report | Apps

Help safeguard press freedom & keep HKFP free for all readers by supporting our team

contribute to hkfp methods
hkfp flask store
YouTube video

Support press freedom & help us surpass 1,000 monthly Patrons: 100% independent, governed by an ethics code & not-for-profit.

Success! You're on the list.

Ho Long Sze Kelly is a Hong Kong-based journalist covering politics, criminal justice, human rights, social welfare and education. As a Senior Reporter at Hong Kong Free Press, she has covered the aftermath of the 2019 extradition bill protests and the Covid-19 pandemic extensively, as well as documented the transformation of her home city under the Beijing-imposed national security law.

Kelly has a bachelor's degree in Journalism from the University of Hong Kong, with a second major in Politics and Public Administration. Prior to joining HKFP in 2020, she was on the frontlines covering the 2019 citywide unrest for South China Morning Post’s Young Post. She also covered sports and youth-related issues.