Coordinated AI disinformation network targets Prime Minister Lawrence Wong

An extensive AI-driven disinformation network is targeting PM Lawrence Wong and Singapore’s stability. Originally uncovered by Lianhe Zaobao, the operation uses deepfakes and SEO poisoning to spread falsehoods about leadership infighting and economic failure.

YoutubeAI.jpg
Thumbnail in disinformation campaign depicts Senior Minister (SM) Lee Hsien Loong and PM Lawrence Wong with text claiming "Lee Hsien Loong hands over to Lawrence Wong! Singapore suddenly shows 'signs of change'! Policy U-turn for peace and internal crisis
AI-Generated Summary
  • A network of over 30 YouTube channels has published hundreds of AI-generated videos spreading fabricated claims about Singapore’s Prime Minister and domestic politics.
  • The operation uses industrialised AI tools to produce long-form content for as little as US$1 to US$2 per video.
  • False narratives include alleged power struggles between PM Lawrence Wong and SM Lee Hsien Loong, alongside debunked claims of a collapsing shipping industry.
Comments
Google News

An extensive disinformation campaign involving hundreds of AI-generated, Chinese-language YouTube videos has been identified targeting Singapore and Prime Minister (PM) Lawrence Wong.

The existence of the operation was first reported by the Chinese-language newspaper Lianhe Zaobao, with subsequent investigations by Channel NewsAsia (CNA) revealing the depth of the coordinated activity.

The campaign has produced nearly 300 videos surfaced over a period of several weeks. Analysis shows that seven in 10 videos specifically target PM Wong, fabricating narratives about his leadership role being under threat.

These videos have accumulated millions of views since they first appeared in late 2024.

A distinctive trait of the operation is the aggressive use of search engine optimisation (SEO) "poisoning". By using hashtags referencing Singapore or PM Wong in both traditional and simplified Chinese, the operators attempt to contaminate search results. This tactic draws in viewers who might not otherwise engage with such disinformation.

Technical analysis suggests at least 10 of the 30 identified YouTube channels act in concert.

One cluster of channels, using the naming prefix "Southern Master", was created within a 20-minute window on 18 January 2026. These accounts published videos at identical intervals, a hallmark of a centrally controlled influence operation.

According to Lin Jiayi, a scholar at Black Dot Research, the videos are likely authored by Chinese large language models (LLMs) like DeepSeek or Ernie Bot. These models follow specific templates that begin with an artificial sense of crisis and conclude with warnings of impending disaster or political upheaval.

The production of this content is highly cost-effective. Magellan Technical Research Institute (MTRI) scientist Sun Yiqun estimated that a 20-minute video costs only US$1 to US$2 to generate. This covers AI-scripting, text-to-speech voiceovers, and synthetic imagery, allowing the operators to profit through traffic monetisation.

One prominent recurring element is the use of a deepfake avatar of the late American investor Charlie Munger. The AI-generated likeness appeared in videos across more than half of the investigated channels. Other techniques include the use of identical scripts and voiceovers paired with different visual templates.

Among the false narratives pushed was a conspiracy theory claiming PM Wong was to be sacked by Senior Minister (SM) Lee Hsien Loong.

Video titles referred to PM Wong as the "shortest-serving PM" and declared he had "fallen from power". No factual evidence exists to support these claims.

The campaign also targeted Singapore's economic standing, claiming the trade hub was threatened by a Hainan port and citing a "mass exodus" of Fortune 500 companies. However, official data shows the Port of Singapore handled a record 44.66 million containers in 2025, remaining the world's second-busiest container port.

Associate Professor (Assoc Prof) Chong Ja Ian from the National University of Singapore (NUS) noted that the scale suggests a state backer or a well-funded private entity. Assoc Prof Chong stated the campaign could be aimed at causing domestic uncertainty or shaking confidence among the global Chinese-speaking community.

The Ministry of Digital Development and Information (MDDI) stated it was aware of the accounts and their "wholly fabricated and outlandish claims". A spokesperson emphasised that while the government monitors such risks, a discerning public is the most crucial layer of defence against misinformation.

The campaign is part of a broader global trend. Similar coordinated networks have targeted US President Donald Trump and Japanese Prime Minister Sanae Takaichi. In Japan, thousands of social media accounts were found spreading fabrications about Takaichi's government ahead of election campaigning.

Associate Professor (Adjunct) Carol Soon, Vice Chair of the Media Literacy Council, noted that these videos aim to erode institutional trust. Dr Soon warned that as technology becomes more sophisticated, synthetic voices and visuals will become increasingly difficult for the average consumer to distinguish from reality.

Benjamin Ang, a Senior Fellow at the S. Rajaratnam School of International Studies (RSIS), highlighted the need for mother-tongue fact-checking. Mr Ang noted that when narratives appeal to identity in a specific language, English-language debunking might not reach or resonate with the affected audience.

While Google terminated several flagged accounts within 12 hours of being notified, the investigation found that fresh videos and channels often surface quickly.

Experts continue to stress the importance of media and civic literacy as the most effective long-term response to increasingly sophisticated synthetic media.

Related Tags

Share This

Support independent citizen media on Patreon