Translation. Region: Russian Federation –
Source: People's Republic of China in Russian – People's Republic of China in Russian –
An important disclaimer is at the bottom of this article.
Source: People's Republic of China – State Council News
BEIJING, November 15 (Xinhua) — China's Cyberspace Administration has launched a crackdown on deepfakes posing as influencers during e-commerce livestreams, as the country has been hit by a surge in AI-powered fraud.
As the department announced on Friday, authorities recently "severely punished" several online accounts that used AI to impersonate internet celebrities to promote products in live streams and short videos, misleading users. Online platforms initiated a purge of such content, which has so far resulted in the removal of over 8,700 posts and the processing of over 11,000 accounts posing as influencers.
This campaign began after a series of high-profile cases, including one involving actress Wen Zhengrong. Users discovered AI-generated "clones" of her advertising various products in several live broadcasts. The actress herself, who tried to stop the deception, was blocked from chat after asking, "If you are Wen Zhengrong, then who am I?"
Her experience highlights a growing problem in China's rapidly expanding e-commerce sector. As artificial intelligence tools become increasingly accessible, unscrupulous sellers are using them to create sophisticated but fake ads featuring influencers and celebrities to promote sometimes substandard or counterfeit products.
One scheme to promote "deep-sea polyene fish oil," which turned out to be ordinary candy, used a deepfake impersonator of TV presenter Li Zimeng, and Olympic champions Quan Hongchan and Sun Yingsha were impersonated to sell eggs.
"This practice violates the Civil Code of the People's Republic of China," noted Liu Hongchun, executive director of the Human Rights Law Research Center at Yunnan University. "By doing so, the scammers infringe on third-party image rights and violate consumers' right to information."
Although China introduced the “Labeling Measures for Content Generated and Compiled by Artificial Intelligence” on September 1, enforcement in this area still faces certain challenges.
Violators often hide watermarks in inconspicuous corners of videos or use technical means to remove identifiers. Some break the synthesized content into fragments and distribute them across multiple accounts, making them difficult to detect.
Platforms are also facing content moderation challenges. While Douyin, the Chinese version of TikTok, removed over 10,000 deepfake videos featuring Wen Zhengrong and fined 37 accounts, TikTok Vice President Li Liang acknowledged that identifying AI-generated content for copyright infringement remains a "technical challenge for the entire industry."
"The attackers who run such deepfake accounts are constantly using various technical means to counteract the platforms," Li Liang noted in his Weibo post, promising that Douyin would increase investment to address the issue.
For victims, gathering evidence and proving that the content was created by AI is technically complex and time-consuming. According to media reports, Wen Zhengrong's team discovered 50 accounts with deepfakes of the actress within 24 hours, but some of them quickly reappeared in new forms.
Legal experts and industry insiders say tackling this problem requires a comprehensive approach: tougher penalties for violators, improved detection technologies, and greater accountability for platforms.
The Cyberspace Administration of China said it would maintain a “high-pressure” stance, continuing to hold platforms accountable and “expose and remove” malicious marketing accounts.
“We must maintain an open and inclusive attitude toward AI, but we must also adhere to the core principles and ‘red lines’ in its use to truly make the technology a driving force for good and human well-being,” said Liu Xiaonan, a professor at the China University of Political Science and Law. -0-
Please note: This information is raw content obtained directly from the source. It represents an accurate account of the source's assertions and does not necessarily reflect the position of MIL-OSI or its clients.