By Jia Feimao | 2025-10-06

Ominous and pessimistic remarks by Taiwan's President Lai Ching-te turned out to be just artificial intelligence (AI) deepfakes.

"Time is running out for Taiwan. China is getting stronger, and the US can no longer contain it. One of these days, they might just trade us away."

In a video posted on the now-defunct YouTube channel AI Community of Common Destiny, Lai appears to be speaking to Vice President Hsiao Bi-khim.

A closer look reveals that the channel's videos were all fabricated through deepfake technology, altering the words of Taiwan's political leaders. One short clip uploaded on September 17, less than 50 seconds long, had a computer-generated Lai saying: "Even if we buy all of America's [military] equipment, we can't defeat Communist China," "We should use this time to make as much profit as we can" and "As soon as we sense something is wrong, we'll immediately leave Taiwan."

Although the channel claimed the videos were only for entertainment, Lo Tien, adjunct associate professor at National Taiwan University of Science and Technology, told PTS News: "Whether it's voice cloning or facial forgery, the level of realism is quite high. If this infringes upon national interests, then there must be an investigation."

Taiwanese national security officials consider the "pro-China" and "anti-US" narratives in these clips highly misleading and suspect possible involvement by Chinese cyber forces or allied social media operators, Taiwanese media reported.

Because the videos attracted widespread coverage, around the end of September, YouTube took down the AI Community of Common Destiny channel, launched in July.

No harmless prank

Creating fake clips of leaders may look like pranks or clickbait, but leaving them unchecked risks undermining Lai's credibility, Yao-yuan Yeh, director of the Taiwan & East Asia Studies Program at the University of St. Thomas in Houston, Texas, told Focus.

If Taiwan enters a crisis, similar false narratives could be amplified and endanger the public.

"Legally, there needs to be a ban on using AI-generated images or voices of public figures to promote false information, whether it's just for fun or for malicious intent," Yeh added, urging Taiwanese authorities to investigate whether Chinese funding was involved.

While the motives of the AI Community of Common Destiny channel remain unclear, one fact is certain: organizations associated with the Chinese government have already deployed AI-enabled firms to shape public opinion.

In August, engineering professor Brett J. Goldstein and political scientist Brett V. Benson, both from Vanderbilt University in Nashville, Tennessee, revealed in The New York Times that GoLaxy, a firm funded by entities linked to China's military, has been spreading disinformation and conducting malign influence campaigns in Hong Kong and Taiwan.

According to internal documents obtained by the Times, GoLaxy used its so-called Smart Propaganda System (also known as GoPro) to analyze social platform data and build personal profiles, even recommending strategies to erode trust in Taiwan's ruling Democratic Progressive Party.

The company reportedly collected data from more than 5,000 social media accounts belonging to Taiwanese so it could map political attitudes toward China.

Meddling in Taiwan and Hong Kong

In the lead-up to the 2024 Taiwanese presidential election, GoLaxy struck with AI-generated deepfakes, while China-aligned groups spread defeatism and false claims of corruption on social media.

In 2020, it used its army of phony accounts to attack opposition to a Chinese national security law wiping out the remaining freedom of expression in Hong Kong.

"It [GoLaxy] is deeply tied into the Chinese government's security apparatus and the military. They are building new tools that are proposing to do a better job at information operations," James Mulvenon, chief intelligence officer at Pamir Consulting, told the Times.

AI-generated content is particularly dangerous because it can spread "compelling" messages without attracting resistance, said Goldstein and Benson.

"It operates continually, shaping opinion and corroding democratic institutions beneath the surface," they said.

Propaganda has evolved from traditional mass broadcasting to precision targeting based on individual traits, say Taiwanese security officials. To defend against hostile influence, they say, Taiwan must build multilayered defenses spanning technology, legislation, media literacy and civic education.