BEIJING, Nov. 14 (Xinhua) -- Chinese cyberspace authorities are cracking down on the use of artificial intelligence (AI)-generated deepfakes to impersonate public figures in livestreaming e-commerce, as the country grapples with a surge of AI-fuelled fraud.

The Cyberspace Administration of China (CAC) said Friday that authorities had recently "severely dealt with" a batch of online accounts that used AI to mimic celebrities to promote products in live-streams and short videos, misleading users. Online platforms launched a cleanup, which has so far removed over 8,700 pieces of non-compliant content and dealt with more than 11,000 accounts that impersonated public figures.

The campaign follows a series of high-profile cases, including one involving actress Wen Zhengrong, who discovered multiple AI-generated clones of herself promoting products across different live streams. In an attempt to stop the impersonation, Wen confronted one impersonator in a live chat, only to be blocked after asking, "If you are Wen Zhengrong, then who am I?"

Her experience underscores a growing challenge in China's booming e-commerce sector. As AI tools become more accessible and affordable, they have been co-opted by unscrupulous merchants to create sophisticated but fake endorsements from public figures, sometimes for low-quality or counterfeit products.

In one scheme, a forged AI likeness of TV host Li Zimeng was used to promote "deep-sea fish oil," which turned out to be ordinary candy. Olympic champions like Quan Hongchan and Sun Yingsha have also been impersonated to sell eggs.

"These behaviors violate the Civil Code," said Liu Hongchun, an associate professor at the Law School, Yunnan University. "They infringe upon these public figures' portrait rights, and also violate consumers' right to information."

While China implemented the "Artificial Intelligence Generated Content Labeling Measures" on Sept. 1, requiring AI-generated content to be clearly marked, enforcement remains challenging.

Violators often hide watermarks in inconspicuous corners of videos or use technical means to remove identifiers. Some break synthesized content into fragments and distribute them across multiple accounts, making detection difficult.

Platforms also struggle with content moderation. While Douyin, TikTok's Chinese version, has taken down over 10,000 videos impersonating Wen Zhengrong and penalized 37 accounts, its vice president Li Liang acknowledged that identifying AI-generated content for copyright infringement remains an "industry-wide technical challenge."

"Malicious impersonation accounts continuously engage in technical countermeasures against platforms," Li said in a Weibo post, pledging that the platform would increase investment to meet these challenges.

For victims, gathering evidence and proving that the content is AI-generated is technically complex and time-consuming. Wen's team, according to media reports, once reported 50 impersonating accounts in a single day, only to find that some resurfaced quickly in new forms.

Legal experts and industry insiders say addressing the problem requires a multi-pronged approach: stronger penalties for violators, better detection technology, and clearer accountability for platforms.

The CAC said cyberspace authorities would maintain a "high-pressure" stance, continuing to hold platforms accountable and to "dispose of and expose" malicious marketing accounts.

"We should maintain an open and inclusive attitude towards AI, but we must also adhere to bottom lines and red lines in its use -- truly making technology a force for good and for the people's well-being," said Liu Xiaonan, a professor at China University of Political Science and Law. Enditem