Beijing Deployed AI-Generated Deepfakes Targeting Taiwanese President During General Election

Beijing Deployed AI-Generated Deepfakes Targeting Taiwanese President During General Election
Democratic Progressive Party's (DPP) presidential candidate, Lai Ching-te (L), and Taiwan's president, Tsai Ing-wen (R), cheer at the end of the campaign rally in Taipei City, Taiwan on Jan. 11, 2024. (Annabelle Chih/Getty Images)
Raven Wu
Ellen Wan
2/2/2024
Updated:
2/2/2024
0:00
News Analysis

The Chinese Communist Party (CCP) has always tried its best to interfere with Taiwan’s general elections. In addition to the threat of violence, it has also used various long-established infiltration channels to spread false news in an attempt to reshape public opinion.

In the run-up to this year’s election on Jan. 13, the Chinese regime went one step further to take advantage of AI technology in its meddling efforts. Although its underhanded actions were futile, the seriousness of its disinformation campaign was alarming.

In the first week of January, a large number of bot-like accounts posted an e-book called “The Secret History of Tsai Ing-wen” on Facebook, X, YouTube, Reddit, Telegram, and other platforms.

The e-book defamed Ms. Tsai’s family background and private life by presenting a combination of distorted historical events and rumors.

The style of the e-book, from its choice of words to its logic, is very similar to CCP propaganda writing.

Although Ms. Tsai is about to leave office, attacking her was another opportunity for the CCP to alienate voters from her Democratic Progressive Party (DPP), which refuses to acknowledge the so-called “One-China framework” defined in 1992 as the basis for the official diplomatic relations with mainland China.

A Taiwanese national security official told the Central News Agency on Jan. 10 that there has been a proliferation of AI-generated deepfake videos on the Internet since the beginning of the year. These videos feature a virtual anchor reading content from the e-book “The Secret History of Tsai Ing-wen,” and have been distributed through hundreds of fake accounts on YouTube.

A detailed assessment revealed that these activities are closely linked to China’s Ministry of State Security, the official said.

He further explained that AI technology was used to create videos in which virtual anchors read various attack narratives from the e-book. These videos then rapidly spread across various social media platforms. They first appeared on YouTube, posted by over 100 different accounts.

The content is the same, but the AI-generated anchors are different, with hundreds of other accounts posting the same content.

Multiple Facebook accounts also joined the relay of this disinformation war.

Taiwan’s national security agency discovered that the video was created using an app developed by ByteDance, TikTok’s parent company. The only difference between the different accounts is that the AI anchors have images of cartoon characters, faces of people from different countries, Santa Claus, and so on. Generally speaking, these are all common tactics of the CCP’s National Security Bureau.

“The world must beware of the CCP, which is adept at weaponizing whatever it can use to attack the free world,” said Japan-based electrical engineer Li Jixin.

“To begin with, the CCP is skilled in producing and disseminating fake news. Now, the emergence of AI technology helped to amply its capability in waging a disinformation war,” he said.

Mr. Li warned that the CCP’s meddling in Taiwan’s election could be a rehearsal for its next move to interfere with the U.S. presidential election in November.

“The CCP knows that in the short period leading up to an important election, the public is unable to verify whether the information they receive is true or false,” he said.

“Combined with the atmosphere of panic it has created through military provocations, the CCP can easily manipulate voters during important elections in Taiwan and in the United States.”