China's Communist Party's AI video and fake news introduce the election. He urges "beware of fear": controversial messages soared to 2.5 million

With the rapid development of artificial intelligence (AI) technology, the impact of deep fake videos and false information on elections has attracted attention. Democratic Progressive Party legislator Lin Junxian pointed out that recently, a large amount of fake information has been poured into candidate websites, and deepfake videos and videos that impersonate candidates’ voices and fabricate speeches have been circulated on social platforms, which may interfere with voters’ judgment. He called on the Ministry of Digital Development to quickly develop AI image labeling specifications and require users to proactively disclose AI-generated content. Platforms should also establish labeling and reporting mechanisms to prevent false information from becoming a tool that affects democracy.

Lin Junxian posted on Facebook that with the advancement of AI technology, fake videos and sounds can now be produced in just a few minutes, but it often takes several days for the parties involved to clarify. He pointed out that recently, a large amount of fake information has been poured into candidate websites, and deepfake videos that impersonate candidates’ voices and fabricate speeches have also appeared on social platforms. Such content will not only directly interfere with voters’ judgment, but is also a common method of “introduction” operations.

Citing a report recently submitted by the National Security Bureau, Lin Junxian said that from January to September this year, more than 2.5 million controversial messages were reported, an increase of nearly 50% compared with the same period last year, and the proportion on Facebook and Threads platforms also increased significantly.

He said that the CCP continues to generate video manipulation information through fake accounts, content farms and AI. As the election is getting closer, the government should be "cautious and fearful."

Regarding the increasingly common situation of AI-generated content and deepfake images, Lin Junxian pointed out that social platforms such as Facebook and Threads currently have automatic detection and labeling functions, but the system may still miss detection, and some people may deliberately avoid detection. Therefore, it may be difficult to reassure the public if we only rely on the platform to make its own judgments.

Lin Junxian believes that the Digital Development Department should quickly develop AI image labeling specifications and establish a more complete identification and disclosure mechanism. He advocated that when users upload AI-generated content, they should proactively disclose it; after the platform detects AI content, it should also add relevant labels.

In addition, regarding AI-generated content that the system failed to successfully identify, Lin Junxian said that the government, the parties involved in the fraud, and general users should have appropriate reporting channels to prevent the continued circulation of relevant content.

Lin Junxian finally emphasized that elections should be a process that allows people to compare political opinions and make choices. If AI is used to fabricate content and manipulate public opinion, it will harm everyone’s judgment. He called on fake videos, fake photos and fake accounts to not be used as tools to influence democracy.