China to regulate the use of generative AI

The Cyberspace Administration of China published a draft of rules aimed at regulating generative artificial intelligence algorithms.

The proposed measures are intended to “facilitate healthy development and standardisation of generative AI.” Meanwhile, authorities said they support “the independence of algorithms.”

According to the document, service providers must guarantee:

• content accuracy;
• respect for intellectual property;
• non-discrimination;
• ensuring national security.

The document separately notes that generative models are forbidden from inciting the overthrow of the state.

“The AI-generated content must reflect the core values of socialism and must not call for the overthrow of the government and the socialist system.”

The rules place responsibility for AI-generated text, images and audio on service providers. Before the public launch of services, authorities must assess their safety.

Additionally, the document bars AI tools from disclosing or using citizens’ personal data without their consent. The same applies to trade secrets.

The draft is open for public consultation, which will run until 5 May 2023.

In February, reports said that Chinese authorities ordered local tech giants not to provide access to ChatGPT due to the impossibility of censoring responses.

In January, China introduced rules, prohibiting the creation of deepfakes without the subject’s consent or in ways that threaten national security.