China Issues Security Requirements For Generative AI Services: Blacklist Of Data Sources

JAKARTA - China has issued proposed security requirements for companies offering services supported by generative artificial intelligence, including blacklists of sources that should not be used to train AI models.

Generative AI, which is popular thanks to the successful chatbot ChatGPT from OpenAI, learned how to take action from past data and create new content such as text or images based on the training.

The requirement was issued Wednesday 11 October by the National Information Security Standardization Committee, involving officials from China's Cyber Administration (CAC), the Ministry of Industry and Information Technology, and the police.

The committee proposes conducting a safety assessment of any content used to train a publicly displayed generative AI model, with "more than 5% of illegal and malicious information" to be blacklisted.

The information includes "proposing terrorism" or violence, as well as "overthrowing the socialist system", "corruptioning the country's image", and "disrupting national unity and social stability".

The regulation also states that censored information on the Chinese internet should not be used to train models.

Its issuance comes just over a month after regulators allowed several Chinese technology companies, including search engine giant Baidu, to launch their AI-generatively supported chatbots to the public.

CAC has since April said it wants companies to submit security assessments to authorities before launching a generating AI-backed service to the public.

In July, cyber-prone regulators published measures governing the service analysts said were much lighter than the measures outlined in the April draft.

Draft security requirements published last Wednesday require organizations training these AI models seeking individual approvals that use their personal information, including biometric data, for training purposes. They also detailed guidelines on how to avoid intellectual property rights violations.

Countries around the world are struggling to set rules for this technology. China sees AI as an area where it wants to compete with the United States and has set a target to become a world leader in this area by 2030.