China’s First-of-its-kind Rules Governing Generative Artificial Intelligence to Become Effective in August

Credit: Visual China

BEIJING，July 14 (TiPost) — China’s first regulation governing generative AI services was released on Thursday by the Cyberspace Administration of China (CAC) and several other regulators. The new rule will come into effect on Aug. 15.

The Measures for the Administration of Generative Artificial Intelligence (the Measures) makes it clear that China emphasizes on both development and security and promotes innovation and governance in accordance with the law. Effective ways will be adopted to encourage the innovative development of generative AI, and the country will implement inclusive, prudent and categorized oversight of relevant services.

According to Cai Peng, a partner of Zhonglun Law Firm, the Measures reflects China’s encouraging and supportive attitude towards the development of AI, and aims to create a safe and relaxed environment for relevant service providers.  

Cai pointed out that the Measures puts forward for the first time the categorized and graded regulation of AI-related services given their difference from other industries and fields, and related guidelines is expected to be introduced to provide clear guidance for the development of such technologies and management of services.

Consistent with the Draft rules published in April, the Measures strictly sets the security assessment and algorithm filing as pre-conditions for an AIGC product launch, as specified in Article 17.

Article 23 of the Measures states that the provider of generative AI services shall obtain the licenses in accordance with the law; foreign investment in generative AI services shall be in accordance with the provisions of laws and administrative regulations relating to foreign investment.

It is also added in the Measures that the providers should cooperate with the relevant authorities in the supervision and inspection of generative artificial intelligence services in according with law.

Article 8 of the Measures states that when manual labeling of data is adopted in the development of generative artificial intelligence products, the provider shall formulate clear, specific, and operable labeling rules that meet the requirements of these Measures, conduct necessary training for labeling personnel, and verify the correctness of labeling contents by sampling.

It is understood that many large and small companies that have launched AI big models are applying for algorithm filing and security assessment to CAC, waiting to pass the audit before commercializing their big models.

The emergence of ChatGPT has triggered a wave of generative artificial intelligence (AIGC) products, and concerns associated with how to regulate relevant services have been raised.

Governments and authorities around the world are racing to legislate against potential abuses of the technology. Both the U.S. federal government and its legislature have clearly stated their concern for AI, with unprecedented consensus.

The EU Parliament announced on June 18 that its AI Act has entered the final discussion stage, which will be the world’s first comprehensive legislation governing the technology once approved.

Overall, the underlying logic of global regulation of AI is basically the same: control the harm of AI and support its development. It is the mission of global regulators to race against Al and control it before it goes off the rails.

更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App

​Read More