China Releases Draft Measures for Managing Generative AI Services

China Releases Draft Measures for Managing Generative AI Services

Chinese regulators have released draft rules designed to manage the development of generative AI products like ChatGPT from OpenAI.

The draft, which was released on Tuesday, demands that companies submit security assessments to authorities before launching their offerings to the public.

Named the “Administrative Measures for Generative Artificial Intelligence Services,” the proposed legislation is intended to guarantee the proper growth and regulated implementation of generative AI technology, as stated in the draft law.

Also Read: Will Pause on GPT Development Hinder China’s AI Progress?

Draft regulations by the Cyberspace Administration of China (CAC) have come at a time when various governments are finding ways to deal with the technology.

Many tech giants are embracing AI, especially chatbots and generative tools, and the risks associated with it, which drew the attention of authorities to prepare the framework.

Socialist values in generative AI standards

This step from regulators also follows recent bids by Chinese tech giants including SenseTime, Alibaba, and Baidu to develop their own AI-powered products to dive into the frenzy of the AI era.

“Before providing services to the public that use generative AI products, a security assessment shall be applied for through national internet regulatory departments,” reads the draft law, released by the CAC.

In order to ensure compliance with the country’s core socialist values, the CAC has stated that generative AI content must meet certain standards.

Providers of generative AI services are accountable for the legality of the data used to train their products and should take precautions to prevent discriminatory algorithms and training data.

Three months max to update technology

The regulator has mandated that service providers require users to provide their authentic identities and associated information. Providers who fail to adhere to such regulations may face fines, suspension of services, or even criminal investigations.

If inappropriate content is produced by a company’s platform, meanwhile, they must update their technology within three months to prevent a recurrence of similar content.

Until May 10, the public is invited to comment on the proposals with the regulations projected to take effect at some point this year.

AI-powered weather forecast

AI technology is expanding its sphere of influence each day, and weather forecasts have made it onto the list.

Chinese scientists have developed an advanced AI weather forecasting system called FengWu, capable of producing highly accurate global medium-range weather forecasts for up to 10 days within 30 seconds.

The system utilizes deep learning architecture and multi-model and multi-task approaches to generate precise predictions.

Compared to traditional physical models, FengWu reduces error rates by 19.4% and extends the forecast range to 10.75 days based on reanalysis data.

This technology could be used to support industries such as agriculture and forestry, and could supplement traditional physical models in weather forecast digitalization.

Image credits: Shutterstock, CC images, Midjourney, Unsplash.