Internet companies Illustration: VCG

 
The Cyberspace Administration of China (GAC) is mulling regulations for deep synthesis technology in China, in order to protect people's legitimate rights and interests. 

The draft, which was posted on the official website of the GAC on Friday, said that deep synthesis service providers should strengthen training data management, ensure that data processing is legal and proper, and take necessary measures to ensure data security. 

If training data contains data involving personal information, it should also comply with the relevant regulations on the protection of personal information, and personal information should not be processed illegally, the draft said.

Where a deep synthesis service provider offers significant editing functions for biometric information such as the face and human voice, consent should be sought before editing of the information, the draft explained.

The draft set the deadline for feedback as February 28, 2022.

The purpose of the document is to guide this technology onto a healthy development track, prevent risks, and promote its positive role in tech application, Zuo Xiaodong, vice president of the China Information Security Research Institute, told the Global Times on Friday. 

According to Zuo, deep synthesis is a typical application in artificial intelligence technology. It used to be commonly known as "deep-fake" because it could be used to fake audio and video content that didn't exist, which poses potential challenges for national security. 

But this technology has many positive applications, such as film and television production, digital fitting, and helping the voiceless. In this case the technology does not use "deepfake," but a neutral "deep synthesis," he explained. 

China is set to strengthen security assessment of new internet-related technologies and applications using voice-based social network and deep-fake technology. 

In March last year, Chinese regulators summoned a total of 11 heavyweight Chinese internet companies including Tencent, Xiaomi and Kuaishou, in order to strengthen security assessment of voice-based software and new internet technologies and applications involving deep-fake technology.


The Global Times reported previously that the latest deep-fake example is Avatarify, a face-changing app adopting AI to allow users to replace their own faces with other people's facial graphic for photography or making videos. The app went viral on TikTok in February 2021 and has already had more than 1.5 million downloads, according to media reports. But the app was removed from Apple's China app store on March 2 due to concerns about invasion of privacy.

Local police in Zhengzhou, Central China's Henan Province warned in March 2021 that face-changing software has certain security risks, such as personal information being leaked, and it is common for such technology to be used in illegal and criminal fields, and the videos produced are often suspected of being pornographic and obscene.
 
China has been making efforts to enhance data security. The country has issued regulations targeting algorithm recommendations, which will take effect on March 1, in a bid to promote the sustainable development of algorithm services and strengthen supervision to ensure long-term market order.

In January, China's top regulators released a guideline regulating various aspects of online platform businesses, including their investments in financial institutions, big data-enabled price discrimination against customers and monopolistic behavior, in order to promote the industry's sound and sustained development. 

In August last year, China passed its Personal Information Protection Law, marking the latest addition to the regulatory framework that the country has been creating for the fast-growing digital economy.