China has introduced new guidelines requiring generative AI services to adhere to socialist principles, while promoting industrial applications of the technology. The regulations, set to take effect on August 15th, aim to limit public use of generative AI services while encouraging their development for industrial purposes.
The Cyberspace Administration of China (CAC) has revised its initial draft rules, softening its stance. These interim regulations only apply to organizations providing generative AI services to the public, excluding those developing the technology for non-mass-market use.
The guidelines emphasize that generative AI services must align with the core values of socialism and refrain from undermining state power or the socialist system. Notably, the updated rules have eliminated the potential fines mentioned in the April proposal.
China has been actively working to enhance its generative AI capabilities and aims to challenge the US’s current dominance in the field.
However, China faces challenges in this endeavor due to its stringent control over internet access and information flow within the country. The government has instructed its tech giants to refrain from using ChatGPT to prevent “uncensored replies,” despite the tool being unavailable in China. Authorities have also taken measures against individuals using ChatGPT, including the arrest of a man accused of utilizing the chatbot to produce fabricated articles.
Chinese tech giants Alibaba and Baidu are now developing their own generative AI tools. Baidu recently unveiled its chatbot Ernie, although its launch received a lukewarm response from investors.
China’s regulations on generative AI also prioritize the protection of intellectual property rights for training data. They explicitly prohibit the use of algorithms, data, platforms, or any advantages that promote monopoly or unfair competition. All training data must originate from legitimate sources approved by the government. Service providers are obligated to accommodate individuals’ requests for reviewing or correcting information collected for AI models.
The Chinese government has expressed its commitment to fostering the development of generative AI by providing support for infrastructure and public training initiatives.
China is not the only country grappling with the balance between innovation in generative AI and ensuring public safety. The European Union is currently deliberating on its AI Act, while the Biden administration in the United States has outlined plans to support AI development. Additionally, the US Federal Trade Commission recently initiated an investigation into OpenAI, the creator of ChatGPT, over potential consumer harm.
