Chinese regulators are rolling out new regulations to rein in the use of deep synthesis technology as part of the government’s ongoing campaign to clamp down on online rumours and disinformation.
The Administrative Provisions on Deep Synthesis for Internet Information Service, which will take effect on January 10, were co-published on November 25 by the Cyberspace Administration of China, the Ministry of Industry and Information Technology and the Public Security Ministry.
Deep synthesis service providers and users are required to make sure any doctored content using the technology is explicitly labelled and can be traced back to its source. If using the technology to edit someone’s image or voice, the person in question should be notified and their consent obtained, according to the regulations.
When reposting news made by the technology, the source can only be from the government-approved list of news outlets, which was last updated in October with 1,358 names.
Deep synthesis service providers are also urged to abide by local laws, respect ethics, and maintain the “correct political direction and correct public opinion orientation”, according to the new law.
China’s top internet regulator said it was concerned that unchecked development and use of deep synthesis could lead to its use in criminal activities such as online scams or defamation, according to an article published on the cyberspace administration’s WeChat public account.
The administration defined deep synthesis as the use of technologies, including deep learning and augmented reality, to generate text, images, audio, video and create virtual scenes.
Previously, the responsibility for deep synthesis regulation was spread among multiple authorities, but the move to implement a stand-alone law is a reflection of the rapid development of the technology and the regulatory challenges it poses.
One of the most notorious applications of the technology is deepfakes, where synthetic media is used to swap the face or voice of one person for another. One infamous case involved the use of Mark Zuckerberg’s likeness to doctor a video where the Facebook founder appears to brag about controlling stolen data from billions of people.
Deepfakes, which are getting harder to detect due to the advancement of the technology, are used globally to generate celebrity porn videos, produce fake news, and commit financial fraud. Western social media platforms like Twitter and Facebook have introduced specific rules to detect and prevent the spread of deepfake-facilitated disinformation.
Sun Weimin, head of the cybersecurity coordination bureau at the Cyberspace Administration of China. Photo: Weibo
China’s cyberspace watchdog also said it would promote the lawful and rational utilisation of deep synthesis and ensure its healthy development.
“By implementing … laws and regulations and stipulating systematic and bespoke provisions for the use of deep synthesis in online information services … [we] will provide powerful legal protection to ensure and facilitate the orderly development of deep synthesis services,” said the cyberspace administration.
Shi Jianzhong, vice principal at the China University of Political Science and Law, said the development of deep synthesis technology posed both challenges and opportunities for the artificial intelligence industry.
“Developing deep synthesis technology and broadening its use cases will motivate the whole artificial intelligence industry to push the technologies to grow and to be used for good causes,” said Shi.