Chinese artificial intelligence
Tang Ying | Noor Photos | Getty Images
Chinese regulators finalized the first rules governing artificial intelligence generated on Thursday, as the country looks to tighten oversight of the fast-growing technology.
China’s powerful Cyberspace Administration (CAC) said it worked with several other regulators on the new rules, which will come into effect on Aug. 15.
Generative artificial intelligence is a rapidly growing field of technology in which artificial intelligence services generate content such as text or images. ChatGPT, developed by the US company OpenAI, is the most famous example, allowing users to prompt chatbots and receive responses to queries.
These services are trained on large amounts of data. However, the success of ChatGPT has sparked a wave of rival services, raising concerns among regulators around the world about the potential risks of the technology.
Chinese tech giants have also jumped on the bandwagon, announcing plans and launching their own generative AI services.
But China, which tightly controls its domestic internet through censorship and regulation, is watching AI developments closely. Chinese regulators worry that the services may produce content that runs counter to Beijing’s views or ideology.
That’s part of the reason why Chinese tech companies have been cautious about launching services like ChatGPT. Chinese companies have not focused their technology on enterprise and narrow uses, rather than comprehensive services that are widely available to the public.
For example, Alibaba this month launched an artificial intelligence tool called Unified Wanxiang that can generate images based on prompts, but it is only available for beta testing by corporate customers.
However, the CAC’s rules provide a framework for tech giants to collaborate on technology.
What are the rules for generative AI in China?
For example, the rules apply only to generative AI services offered to the public, not those being developed by research institutions.
Generative AI services require a license to operate, the CAC said.
If an AI-generated service provider discovers “illegal” content, it should take steps to stop generating that content, improve its algorithms, and then report the material to the appropriate authorities.
Providers of these services must conduct security assessments of their products to ensure the security of user information.
China’s AI-generated services must also adhere to “socialist core values,” the Cyberspace Administration of China said.
Still, regulators are struggling to strike a balance between making China a leader in artificial intelligence and keeping tabs on its development.
The CAC’s rules say it aims to encourage innovative applications of generative artificial intelligence and support the development of related infrastructure, such as semiconductors.
Svlook