The regulation, jointly launched by the Cyberspace Administration of China (CAC), the country's internet watchdog, and several other authorities, including the National Development and Reform Commission and the Ministry of Science and Technology and the Ministry of Education, will go into effect on Aug. 15 this year.
An artificial intelligence-generated picture is exhibited at the 2023 World Artificial Intelligence Conference, July 6, 2023. [People's Daily Online/Wang Chu]
Generative AI is a technology capable of generating text, images, voices, videos, codes and other content based on algorithms, models and rules. It is a new type of AI system that generates original content through learning massive data sets, which is of great significance for people's well-being, economic development, national security and strategic competition.
A CAC official said that the regulation aims at promoting the sound development of generative AI and its standard applications, safeguarding national security and social interests and protecting the legitimate rights and interests of citizens, legal entities and organizations.
The regulation puts forward a slew of measures to boost innovative development. It encourages innovative applications of generative AI technology in various industries and fields, healthy and positive content with the technology, optimization of application scenarios and a sound ecology for the technology.
It also supports industrial organizations, enterprises, education and scientific research institutions, public cultural institutions and other relevant organizations to carry out collaboration on technological innovation, data resource construction, commercialization and risk prevention of generative AI.
The regulation encourages independent innovation in the algorithms, frameworks, chips and supporting software platforms of generative AI. It also calls for equal and mutually beneficial international cooperation and exchanges on the technology, and supports participation in the formulation of international rules related to the technology.
A man chats with ERNIE Bot, an artificial intelligent chatbot service launched by Chinese tech giant Baidu at the 2023 World Artificial Intelligence Conference, July 6, 2023. [People's Daily Online/Long Wei]
The booming of generative AI technology has created new opportunities for economic and social development, but has also brought problems such as the spreading of fake information, the infringement of personal information, data safety issues, as well as bias and discrimination.
Therefore, the new regulation requests generative AI service providers and users to obey laws and rules, and respect social morality and ethics.
It clearly stipulates that service providers and users must not jeopardize national security or interests, harm national image, or incite secession. They must not promote terrorism, extremism, ethnic hatred or ethnic discrimination.
Spreading violence, obscenity, fake and harmful information, and other information prohibited by laws and regulations is also forbidden.
To address the possible risks of illegal collection and usage of personal information, as well as infringement of intellectual property rights (IPR), the regulation makes stipulations on respecting IPR and legitimate rights and interests, and improving the accuracy and reliability of AI-generated content.
Services providers shall respect IPR and business ethics. They must not infringe upon the rights of portrait, reputation, honor or privacy, or the personal information of others.
People visit a section on artificial intelligence content creation at the 2023 World Artificial Intelligence Conference, July 6, 2023. [People's Daily Online/Xu Kunde]
According to the regulation, service providers also need to improve their disposal plans. It is expected to make generative AI services more transparent and prevent the spread of illegal content.
Compared to previous technologies, generative AI has many new features. Based on these features, the regulation better adapts its supervising models to suit innovative development, and encourages relevant departments to adopt more targeted management measures.
Xin Yongfei, director of the Policy and Economic Research Institute, China Academy of Information and Communications Technology, told People's Daily that AI's black box cannot explain the reasons for its decisions to the public due to its different way of perception from people, which will lead to uncontrollable behaviors that make supervision difficult.
Therefore, the interim regulation on the management of generative AI services maintains previous supervising methods and exercises classified and grading supervision. It sets requirements on security assessment, algorithm recordation and information disclosure, so as to strengthen the supervision and inspection by relevant authorities and improve China's AI governance system.
The regulation also made stipulations based on other features of AI.
For instance, taking into consideration the massiveness and heterogeneity of AI training data, it stipulates that service providers shall launch and optimize AI training according to the law. By adopting effective measures, service providers will be able to improve the quality of training data and make the data more authentic, precise, objective and diverse.