China will introduce rules governing the use of deep synthesis technology in January 2023. Deepfakes, which use artificial intelligence to manipulate images and videos, are a concern for Beijing as it tightens its grip on online content.
Photography Link | Stock | Getty Images
In January, China will introduce its first regulations targeting “deep fakes,” tightening its grip on internet content.
Deepfakes are images or videos that are synthetically generated or altered using a form of artificial intelligence. The technology can be used to alter existing videos, such as putting politicians’ faces on top of existing videos, or even creating fake speeches.
The result is fake media that appears to be real, but is not.
Beijing announced rules governing “deep synthesis technology” earlier this year and finalized them in December. They will come into effect on January 1. 10.
Here are some key terms:
- Users must consent if their images are to be used in any deep compositing techniques.
- Deep synthesis services cannot use this technology to spread fake news.
- Deepfake services need to verify the real identity of users.
- Composite content must have some kind of notification informing the user that the image or video has been altered by technology.
- Content that violates existing laws and that threatens national security and interests, damages the country’s image, or disrupts the economy is prohibited.
The powerful State Internet Information Office is the regulator behind the rules.
Since late 2020, China has sought to rein in the power of its domestic tech giants, introducing sweeping regulations in areas ranging from antitrust to data protection. But it also seeks to regulate emerging technologies, and has gone further than any other country in terms of technology rules.
Earlier this year, China introduced a rule governing how tech companies use recommendation algorithms in another unprecedented law.
Analysts say the law addresses two goals — stricter online censorship and regulation that stays ahead of new technologies.
“Chinese authorities are clearly eager to crack down on anti-government elements’ ability to use deepfakes by senior leaders, including Xi Jinping, to spread anti-government claims,” Paul Triolo, head of technology policy at consultancy Albright Stonebridge, told CNBC.
“But the rules also show that Chinese authorities are trying to tackle thorny online content issues in a way few other countries have done, seeking to stay one step ahead when new technologies such as artificial intelligence-generated content begin to proliferate online.”
Triolo added that Beijing’s AI regulations in recent years “are designed to keep content policing and censorship efforts one step ahead of emerging technologies, ensuring that Beijing can continue to anticipate the emergence of technologies that can be used to circumvent AI’s overall control system.”
Deep compositing isn’t all bad. It could have some positive applications in areas such as education and healthcare.
But China is grappling with its negative role in creating disinformation.
Kendra Schaefer, a Beijing-based partner at Trivium China Consulting, pointed CNBC to a note she published when the draft rule was released in February, in which she discussed the impact of the landmark regulation.
“Interestingly, China is targeting one of the serious threats facing modern society: the decline in trust in what we see and hear, and the increasing difficulty of distinguishing truth from lies,” the note said.
Various regulators in China have been accumulating experience in enforcing technical rules through the introduction of regulations. Some parts of the deepfake regulations are unclear, such as how to prove that you have given someone their consent to use their image. But overall, Trivium said in its note that China’s existing regulatory system will help it enforce the rules.
“China is able to set these rules because it already has systems in place to control the transmission of content in online spaces, as well as regulators to enforce them,” the note said.