AI-Generated Chinese Videos Spread Misinformation About Singapore, Exposing Linguistic Vulnerabilities

A wave of sensational Chinese-language videos circulating on social media platforms is falsely claiming Singapore’s political leadership faces “turmoil” and “internal strife.” With provocative titles like “Singapore is starting to bleed” and “The chaos in Singapore,” these misleading videos represent more than mere clickbait—they pose a potential threat to public trust in institutions and the broader economy.

Researchers have uncovered a low-cost production system behind this surge of content. Bad actors are leveraging generative AI tools such as DeepSeek and Ernie to automate the entire production process—from scriptwriting and voice-overs to video editing with captions—for as little as US$1 to US$2 per 20-minute video. This technological efficiency has enabled YouTube and TikTok channels to produce hundreds of videos in recent months that spread misleading narratives about regional politics.

Analysis of these channels reveals a diverse collection of content sharing common characteristics. Some recycle stock footage or old TV clips with rapid voice-overs and captions, while others feature individual influencers speaking directly to viewers. Despite different formats, they frequently share identical scripts verbatim. While not explicitly illegal, the content is consistently sensational and misleading.

Creators’ motivations vary from political agendas to advertising revenue, with some potentially building loyal audiences for future financial scams.

Singapore’s vulnerability to this misinformation stems from a significant linguistic blind spot seen in many countries: while English-language content receives robust moderation, vernacular content remains under-monitored. Research from the Harvard Kennedy School Misinformation Review confirms that most disinformation monitoring and debunking occurs in languages of high-income Western countries, leaving Singapore’s other official languages—Mandarin, Malay, and Tamil—relatively unprotected.

This linguistic gap represents a structural weakness in national security. With most experts primarily monitoring information in English, early warning signs of hostile information campaigns circulating in other linguistic communities go undetected.

Global tech platforms like Meta struggle with this problem, as their automated filters are notably less effective at catching nuances, slang, and cultural context in non-English content. Reports show these platforms often fail to stop dangerous disinformation even in widely spoken languages due to insufficient localized linguistic expertise to distinguish between legitimate political discourse and coordinated inauthentic behavior.

The issue extends beyond language to social divisions in Singapore’s multicultural society. According to the 2020 Population Census, while nearly half of residents primarily speak English at home, usage of mother tongue languages varies significantly based on age, education level, socioeconomic status, and immigration background.

Studies by the Institute of Policy Studies demonstrate that language proficiency shapes both identity and news consumption patterns. When residents who rely primarily on Mandarin, Malay, or Tamil sources are underserved by fact-checking resources, different population segments may develop drastically different understandings of events based on their linguistic digital environment—ultimately undermining the national consensus required for social stability.

Even when malicious content is detected, effective response must overcome the language barrier. When a fake Chinese-language video goes viral, official debunking limited to English-language media creates a psychological disconnect. If consumers’ primary information comes from sensational non-English content, but corrections appear only in English, the debunking may never reach its intended audience.

This disconnect can be exploited by bad actors who frame official responses as “government suppression” of “truths” that only mother tongue language audiences are “brave enough” to hear, potentially fueling conspiracy narratives about “English-speaking elites” versus “neglected” vernacular speakers.

To strengthen Singapore’s digital defenses against these threats, experts recommend expanding multilingual monitoring and fact-checking capabilities. This requires developing specialists with cultural intelligence who understand how narratives affect different groups, alongside deploying Asia-developed multilingual AI models to detect sensational non-English content before it becomes viral.

Community influencers and opinion leaders proficient in mother tongue languages can play crucial roles by debunking misinformation in the same languages and on the same platforms—like WhatsApp, WeChat, or Telegram—where it originates. These trusted voices can help communities understand the motives behind misleading videos and develop critical information consumption skills regardless of language.

Singapore’s resilience in the digital age ultimately depends on maintaining a common understanding of facts that transcends race, language, education, and social status. By strengthening multilingual defenses, the nation can transform what hostile actors see as a potential weakness into a unified strength.