Taiwan has flagged nearly 2,700 social media accounts to Meta, alleging that they are part of a coordinated China-linked influence campaign targeting public opinion ahead of the island’s November 28, 2026 local elections. The accounts were identified primarily on Threads, Meta’s text-based social media platform, which has a substantial user base in Taiwan. Taiwanese officials believe the activity was designed to amplify political divisions, discourage voting and weaken public confidence in democratic institutions.

The allegations come from three Taiwanese government officials and internal reports reviewed by Reuters. According to the reports, most of the accounts were created in the previous four months and appeared to have been registered in coordinated batches. Taiwan also identified approximately 260 accounts that circulated more than 2,150 AI-generated videos targeting politicians and policies associated with the ruling Democratic Progressive Party (DPP). Meta said it had removed the accounts reported by Taiwan, but stated that none had posted election-related content and most had no followers. China has not responded to the latest request for comment and has previously rejected accusations of election interference.

Key Takeaways

  • Taiwan flagged nearly 2,700 suspected China-linked accounts to Meta ahead of its November 28 local elections.
  • Most of the accounts were reportedly discovered on Meta’s Threads platform.
  • Nearly all the flagged accounts were created within the previous four months.
  • Of 2,400 accounts detailed in one government report, 74% listed China as their location.
  • Approximately 260 accounts allegedly circulated more than 2,150 AI-generated videos.
  • The videos reportedly attracted 2.35 million interactions, including comments and shares.
  • Meta said it removed the flagged accounts, many of which had already been disabled by its systems.
  • Taiwan believes the campaign sought to deepen political divisions and discourage voter participation rather than simply promote a particular candidate.

Taiwan Flags Nearly 2,700 Accounts to Meta

Taiwanese authorities have identified nearly 2,700 social media accounts suspected of participating in an organised influence operation targeting the island’s political environment.

The accounts were reported to Meta, which operates Facebook, Instagram and Threads.

According to Reuters, Taiwanese officials identified the accounts during investigations into suspicious activity on Threads. The platform has become popular in Taiwan, making it an important channel for political discussion and public communication.

One government report examined approximately 2,400 of the flagged accounts and found that:

Account informationReported figure
Threads accounts detailed in one report2,400
Accounts listing China as their location74%
Accounts with no disclosed locationApproximately 25%
Accounts listing Russia1%
Identified account clusters27
Approximate age of most accountsCreated within the previous four months

The accounts reportedly used similar Chinese-language names and different numerical identifiers. Investigators said the pattern suggested that the profiles may have been created in batches.

Taiwanese officials assessed that operators based in China or Russia may have been involved. However, the account-location information and creation patterns are indicators cited by investigators, not independently conclusive proof of who controlled every account.

How the Suspected Influence Campaign Worked

Taiwanese officials believe the suspected campaign used a gradual approach to build credibility before introducing divisive political messages.

Rather than immediately publishing political content, some accounts reportedly began by presenting themselves as ordinary Taiwanese users.

They shared everyday topics, including entertainment, lifestyle discussions and fortune-telling content. This activity could make the profiles appear more authentic and help them blend into ordinary conversations.

According to officials, some accounts subsequently began criticising the government and attacking political candidates.

The messaging allegedly shifted towards expressing disappointment with the ruling party and suggesting that voters should stop participating in elections.

The strategy is significant because influence operations do not necessarily need to persuade people to support a particular political candidate. They can also attempt to weaken confidence in political institutions, increase social divisions or discourage participation.

One Taiwanese security official told Reuters that the suspected operators were trying to make people turn against each other and become disillusioned with democracy.

Taiwan’s assessment is that the campaign sought to influence the wider political environment ahead of the local elections.

AI-Generated Videos Target Taiwanese Politicians

One of the most significant elements of the investigation involves the use of artificial intelligence-generated videos.

Taiwanese officials told Meta that approximately 260 accounts had circulated more than 2,150 AI-generated videos targeting politicians and policies associated with the Democratic Progressive Party.

According to an internal government report, the videos attracted approximately 2.35 million interactions, including comments and shares.

AI content investigationReported figure
Accounts involved in circulating videosApproximately 260
AI-generated videos identifiedMore than 2,150
Reported interactionsApproximately 2.35 million
Main political targetDemocratic Progressive Party
Platform identifiedThreads

The content reportedly focused on the DPP, which is led by Taiwan President Lai Ching-te.

One target was Puma Shen, the party’s candidate for Taipei mayor. Some videos accused Shen of bullying others during his school years. Shen has denied the allegations.

The use of AI-generated videos creates additional challenges for election authorities because fabricated or manipulated content can be produced and distributed quickly.

Such material can also become difficult for ordinary users to distinguish from genuine footage, particularly when it is presented through accounts that appear to belong to local residents.

However, the reported 2.35 million interactions do not establish how many unique people watched the videos or whether the content changed anyone’s political views.

Why Is Taiwan Concerned About the November 28 Elections?

Taiwan’s November 28, 2026 local elections will determine mayors and county chiefs across the island.

The elections are also politically significant because they provide an indication of public support for President Lai Ching-te’s Democratic Progressive Party ahead of the next presidential election, expected in early 2028.

China claims Taiwan as part of its territory, while Taiwan operates with its own democratically elected government. Beijing has repeatedly criticised the DPP and describes it as a separatist political force.

Taipei rejects Beijing’s sovereignty claims.

This long-running political dispute has made Taiwan a frequent target of allegations involving disinformation, cyber operations and foreign influence campaigns.

The latest investigation adds to concerns that social media platforms can be used to amplify political narratives during sensitive election periods.

For Taiwanese authorities, the challenge is not limited to identifying false information. They must also determine whether coordinated networks are attempting to influence how voters perceive candidates, political parties and democratic institutions.

Meta Removes Accounts Flagged by Taiwan

Meta confirmed that it had removed the accounts reported by Taiwan’s government.

The company said many of the profiles had already been disabled by its own systems before the latest reports.

However, Meta disputed the characterisation that the accounts had been actively influencing election-related conversations.

A company spokesperson said none of the flagged accounts had posted about election-related topics and that most had no followers.

Meta described them as low-quality fake accounts and said it would continue taking action against profiles that violate its policies.

This response highlights an important difference between the assessment made by Taiwan’s authorities and the platform’s description of the accounts.

Taiwanese officials believe the accounts formed part of a broader influence operation, while Meta said the flagged profiles had not published election-related posts.

The company did not publicly confirm that the accounts were controlled by the Chinese government.

The distinction matters because the identification of suspicious accounts, their removal from a platform and the attribution of an influence operation to a particular government are separate issues requiring different levels of evidence.

Why Are Influence Operations Moving Towards Threads?

Threads has become an increasingly important platform for public discussion in Taiwan.

Its text-based format allows users to publish short posts, respond to political discussions and circulate content through social networks.

According to Taiwanese officials, the platform’s popularity made it an attractive target for the suspected campaign.

Investigators said they had not identified similar groups of accounts on Instagram during this investigation.

The reported activity also demonstrates how coordinated networks can use multiple types of content to build influence.

These methods can include:

  • Creating large numbers of artificial profiles.
  • Using similar names and account patterns.
  • Sharing ordinary lifestyle content to appear authentic.
  • Amplifying divisive political discussions.
  • Distributing manipulated or AI-generated videos.
  • Encouraging distrust in political institutions.
  • Promoting messages that discourage electoral participation.

Social media companies face the challenge of identifying coordinated behaviour without automatically treating legitimate political discussion or criticism as manipulation.

This is particularly important during election periods, when political debate naturally becomes more intense.

China Denies Election Interference Allegations

Taiwan’s allegations come amid longstanding tensions between Beijing and Taipei.

China considers Taiwan part of its territory and has not ruled out the use of force to bring the island under its control. Taiwan’s government rejects Beijing’s sovereignty claims.

Beijing has repeatedly rejected accusations that it interferes in Taiwan’s elections.

In the latest case, China’s Taiwan Affairs Office did not respond to a request for comment from Reuters.

The evidence described in the Taiwanese government reports includes account creation patterns, location information, clusters of profiles and the circulation of AI-generated videos.

These findings have led Taiwanese officials to suspect a coordinated operation linked to China. However, publicly available information does not establish the identity of every operator or independently prove direct government control over all the accounts.

AI Is Making Election Disinformation More Difficult to Detect

The reported use of more than 2,150 AI-generated videos highlights a growing challenge for election authorities and social media platforms.

Generative AI tools can produce realistic images, video and audio with relatively little technical expertise.

This can make it easier to create misleading political material, impersonate public figures or distribute fabricated claims at scale.

AI-generated content can also be adapted for different audiences and languages, allowing operators to target specific communities.

The challenge extends beyond identifying whether a video is artificially generated. Authorities must also assess who created it, how it was distributed, whether accounts were coordinated and whether the content violated platform policies or election laws.

For platforms such as Meta, detecting suspicious networks requires a combination of automated systems, human investigation and cooperation with relevant authorities.

The Taiwan case demonstrates how AI-generated content and coordinated account activity can become intertwined during an election period.

What the Incident Means for Social Media Companies

The latest allegations increase pressure on social media companies to improve their systems for detecting coordinated influence operations.

Platforms have invested in account verification, automated detection, content moderation and threat-intelligence teams. However, organised networks can adapt their methods when suspicious accounts are removed.

A profile may be disabled, while similar accounts continue to operate under different names or identities.

The incident also raises questions about how platforms should respond to information shared by governments.

Cooperation with authorities can help identify suspicious activity, but platforms must independently assess evidence and apply their policies consistently.

For Meta, the case is particularly relevant because Threads is expanding its role as a platform for public conversation. The company must balance user growth, political expression and efforts to prevent coordinated manipulation.

The Bigger Picture

Taiwan’s report highlights how election influence operations are evolving alongside social media and generative AI. Instead of relying only on obvious political propaganda, suspected networks can use ordinary-looking accounts, lifestyle discussions and AI-generated videos to build credibility before introducing divisive narratives. The reported campaign also illustrates that an influence operation may aim to discourage voting or weaken trust in democratic institutions rather than simply promote a particular candidate.

For technology companies, the challenge is becoming more complex. Meta must identify coordinated networks while protecting legitimate political speech, and election authorities must distinguish suspicious activity from independently expressed opinions. The Taiwan investigation also underlines the importance of transparency about account removals, evidence and attribution, particularly when allegations involve governments and international relations.

Looking Ahead

Taiwan’s November 28 local elections will provide an important test of how authorities and social media platforms respond to suspected foreign influence activity. Taiwanese officials are likely to continue monitoring Threads and other online platforms for coordinated accounts, manipulated content and attempts to discourage participation. Meta’s removal of the flagged profiles addresses part of the immediate concern, but the effectiveness of its wider detection systems will remain under scrutiny.

The incident is also likely to strengthen calls for better detection of AI-generated political content and greater cooperation between governments and technology companies. As generative AI becomes easier to use, identifying the origin and coordination of misleading material may become increasingly difficult. The wider challenge for democracies will be protecting election integrity without undermining open political discussion or treating unverified allegations as established facts.

Get the day’s top stories in your inbox

One concise email. No spam, unsubscribe anytime.