China Takes Strict Steps To Regulate Deep Fake Technology

China’s first-ever set of provisions to regulate the use of deep fake technology officially take effect on Tuesday, intending to prevent the misuse of this technology while adapting to the rapid development of the artificial intelligence (AI) industry.

Introduced jointly by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, and the Ministry of Public Security, the regulations stress that synthetic videos and photos made using deep synthesis technology, commonly known as “deep fake,” must be “clearly labeled” to prevent public confusion.

The regulations stressed the responsibilities of deep fake providers as well as their users, including prohibiting illegal acts by using the technology, setting up a review system, and identifying user information.

“The specialized regulation issued by China for deep synthesis services has and will continue to have a far-reaching impact on the healthy development of internet information services,” said Meng Dan, director of the Institute of Information Engineering, Chinese Academy of Sciences.

Deepfake technologies have made rapid leaps in recent years and techniques including speech synthesis, face-changing, and even digital humans can be a dream come true thanks to that. Simply by downloading a face-changing APP, users can easily replace the face of a famous actor with their own in an image or even an entire scene from a movie in just a few seconds.

However, “from now on, just like advertisements in China, people will see a small label in the corner of these videos and images, which is a good thing to avoid confusion and crime. This shows that China is doing its best to keep up with the incredible development of the internet and the AI industry,” Ding Daoshi, a veteran analyst in the AI sector, told the Global Times.

Other than deep fake, AI art and AI robots are also covered in the new regulations. Lots of problems have occurred before when it comes to face-changing technology. In 2019, ZAO, an AI face-changing application, went viral after people began to upload photos of themselves as the protagonist of various films.

However, three days after the app was released, the company behind ZAO was investigated by the Ministry of Industry and Information Technology for allegedly collecting facial data contrary to laws and regulations and the app was taken down. In addition to such apps, many research institutes and companies are also conducting research on deep synthesis technology.

New rules went into effect in China on Tuesday to regulate the use of deep fakes, the increasingly realistic digital video manipulations that have sparked disinformation fears around the globe. Deepfake technology allows users to replace one person’s face with another in a video, or to put words into a speaker’s mouth, with at times disconcerting realism.

The technique relies on artificial intelligence and has proven popular on social media, where amusing and often uncanny creations featuring face-swapping celebrities abound. However, the technology can “also be used by unscrupulous people to disseminate illegal information defame and sully the reputation of others, and steal identities to commit fraud”, the Chinese cyberspace administration warned last month.

Deepfakes present a “danger to national security and social stability” if they are not regulated, it said. The new regulations require businesses to offer deep fake services to obtain the real identities of their users. They also require deep fake content to be appropriately labeled to avoid “any confusion” on the part of the public.

China has been quick to regulate technologies seen as potential threats to the stability or the power of its Communist Party. Several homegrown tech giants have been forced to hand over details about their algorithms usually closely guarded corporate secrets to authorities in the past year as Beijing asserts control over the sector.

“What China’s top cyberspace regulators are doing now is to restrain bad behavior and to prevent criminal acts. Only by stopping the wrongdoings can those good companies seek further development and avoid unnecessary disputes,” Ding said. At present, there have not been many achievements in terms of specific legal or regulatory restrictions, even in some European countries and the US, which are a step ahead of the rest of the world in AI development.

In 2020, the European Union published the White Paper on Artificial Intelligence, proposing a series of measures for AI R&D and supervision. The white paper also provoked responses from many huge tech companies including Facebook and Google, who feared that they would be targeted as they were after the publishing of the General Data Protection Regulation.

Many countries are also tracking and punishing AI-related crimes through existing laws and regulations concerning social media or data protection. China’s new regulations are also based on the framework of China’s existing laws, noted cyberspace experts.

“Although specific punishment measures have not yet been released, these regulations are formulated following multiple legal provisions, covering areas including network security, e-commerce, data security, and personal information protection,” noted Ding. Therefore, penalties will be imposed following the laws of the relevant field.

“China has formulated regulations promptly based on the development of the internet and the AI market. As the promulgation of regulations synchronizes with the development of the internet and AI, which shows the forward-looking nature of China’s AI development,” Ding said

Authorities also moved against the country’s vibrant gaming sector in August, freezing the approval of new titles and introducing a cap on the amount of time children could spend playing games. And in one of the most visible examples of China’s tech sector crackdown, regulators pulled the plug on what would have been the world’s biggest-ever IPO of fintech giant Ant Group in 2020 just days after its founder Jack Ma criticized local regulators.

In early 2022, a team from the University of Science and Technology of China was awarded for their deep fake video generation methods at the China Artificial Intelligence Competition. In response to concerns that the new regulations may dampen the rapid expansion experienced by the AI market, experts say that the regulations will not hinder the progress of China’s technology industries.