GERMANY-INTERNET-AI-ARTIFICIAL-INTELLIGENCE
Taiwan's National Security Bureau indicated that it tracked approximately 2.5 million pieces of information related to China's controversial relations with Taiwan between January and September 2026, and reminded people to be careful with online information before local elections. The relevant statistics and attribution are based on an assessment released by the National Security Bureau; the image is an illustration of social media and cybersecurity.

Taiwan’s National Security Bureau (NSB) said its national security intelligence teams tracked about 2.5 million items of information related to China-linked messages targeting Taiwan from January to September 2026, up from approximately 1.69 million during the same period last year. The NSB said the content covered energy, food safety, national defense policy and local election issues, and may be intended to influence public opinion and voters’ decisions in Taiwan.

According to an Oct. 20 report by the Central News Agency, the NSB said the messages could be distributed through platforms including Facebook, Threads, YouTube and online forums. Some of the tactics involved the use of abnormal accounts, proxy accounts, content farms, fake websites and AI-generated audio and video to amplify specific narratives.

Data released by the NSB showed that Facebook’s share of the messages rose to 68.9%, from 43.6% during the same period last year, while Threads’ share increased from 4.5% to 9%. The NSB also said government agencies face growing cyberattack threats, warning that the activities could be linked to intelligence gathering and information operations during the election period.

Taiwan will hold local elections in November, with local development, candidates’ platforms, energy and national defense policy among the issues that could become focal points of discussion on social media. The NSB said the information operations could be intended to influence voters’ perceptions of candidates and policies. However, the figures reflect only the number of messages tracked and classified as controversial by the NSB. They cannot be taken to mean that all of the content has been judicially determined to be false, nor can the actual impact of the messages be assessed solely by their volume.

When evaluating online information, voters can first check the publishing account, original documents and news sources. They should avoid forwarding screenshots, edited videos and sensational headlines without clear origins before verifying them. For election-related information, voters can also compare candidates’ official materials, government announcements and reports from multiple credible media outlets.

Information security experts have long warned that generative AI has lowered the barrier to producing realistic images, audio and video, making it more difficult to determine whether content is authentic. How the government releases verifiable evidence, how platforms respond to coordinated behavior, and how the media and public improve media literacy will be key challenges in preserving the quality of public discussion during the election.