TAIPEI, Taiwan – China has been accused of spreading more than 2.5 million pieces of disputed or false online information targeting Taiwan in the first nine months of this year, a dramatic escalation in the gray-zone campaign that the island’s intelligence agency says is aimed at swaying November’s local elections. In a report to lawmakers this week, Taiwan’s National Security Bureau said such information rose by about 48 percent from the more than 1.69 million items recorded during the same period a year earlier. The report, presented to the Legislative Yuan’s committee on foreign and national defense affairs, identified energy policy, food security, defense policy, and the platforms and public images of specific candidates as the main topics of this state-sponsored disinformation. The bureau said Beijing creates narratives through official Chinese media, then amplifies them through organized networks of paid trolls, public relations firms, content farms, AI-generated videos, fake social media accounts, and coordinated comment flooding. The objective, according to the report, is to sow discord among Taiwanese people, undermine trust in democratic institutions, and influence voter preferences ahead of the “nine-in-one” elections scheduled for November, when voters will select mayors, county magistrates, and local councilors across the island. China’s Taiwan Affairs Office did not immediately respond to a request for comment. Beijing, which considers Taiwan part of its territory and has never renounced the use of force, has repeatedly denied any interference in Taiwanese elections or involvement in disinformation operations, calling such accusations groundless attempts to stir up anti-China sentiment. The NSB’s latest figures underline a trend that security officials have been warning about for years: China’s so-called gray-zone tactics — activities below the threshold of conventional war — have expanded beyond military drills and cyberattacks to encompass a sprawling influence apparatus that exploits digital platforms, technological vulnerabilities, and social fault lines. With elections approaching, the bureau characterized the information onslaught as a systematic attempt to shape the political landscape in favor of forces more willing to accommodate Beijing’s sovereignty claims, while discrediting candidates perceived as hostile to Chinese interests.
The National Security Bureau’s assessment provides a rare inside look at the mechanics of a Chinese influence operation. According to the report, the roughly 2.5 million pieces of suspect content were not the product of a single campaign but a sustained, multi-pronged push that grew more intense as election day drew nearer. The report says Beijing’s messaging machinery operates on an assembly-line model: initial storylines are seeded in state media such as the People’s Daily, Xinhua, and CCTV; they are then picked up by overseas Chinese-language outlets and social-media accounts; finally, low-paid workers and automated bots flood comment sections and forums with repetitive pro-Beijing, anti-Taiwan talking points. The content is calibrated to touch on sensitive, high-emotion topics. Energy security, for example, plays on Taiwanese anxieties over possible blackouts and the island’s dependence on imported fuel; food security narratives exploit public disputes over agricultural imports, including pork and eggs; defense policy messaging seeks to portray Taiwan’s military as ineffective, wasteful, or provocative; and candidate-focused content attempts to elevate or destroy reputations with manipulated images and video. The NSB explicitly warned that Beijing is using AI technology on a new level, gathering and analyzing Taiwanese current affairs, candidates’ policies, and polling data to fine-tune its lies and half-truths. “China has expanded the use of AI technology to manipulate information, seeking to influence voter support and election results,” the report reads. This AI-driven approach allows the operation to run at machine speed, identifying emerging conversations and injecting tailored narratives before fact-checkers can respond. Social media platforms named in the report include Facebook, Threads, YouTube, and local online forums, all of which have struggled to curb coordinated inauthentic behavior despite pledging to enforce integrity policies. In addition to overt fake accounts, the bureau said Beijing uses content farms to produce large volumes of blog posts, memes, and short videos that can be shared across platforms. Some content is not overtly political at first but designed to build trust and followers in community groups before shifting to political messaging, a technique known as “sleeper influence.” In many cases, messages are further amplified by real Taiwanese users who share them unknowingly, creating a diffusion network that is almost impossible to quarantine.
Taiwan has repeatedly raised alarms about what it sees as a comprehensive Chinese gray-zone strategy to force Taipei to accept Beijing’s sovereignty claims without triggering a full military invasion. For years, this strategy has manifested as daily military aircraft and naval vessels entering Taiwan’s air defense identification zone, sudden cyberattacks on government and infrastructure websites, and the shrinking of Taiwan’s diplomatic space through Beijing’s pressure on other countries. The disinformation campaign identified by the NSB is the information component of that same strategy. The report itself was submitted to lawmakers as part of an annual briefing on external threats, a routine but critical oversight exercise that allows Taiwan’s intelligence agencies to publicly document patterns of foreign interference. The release of the numbers — 2.5 million items in only nine months — is intended to underscore both the severity and the audacity of Beijing’s activities. China, for its part, has not altered its position. The Taiwan Affairs Office, which is Beijing’s main bureaucratic body overseeing island-related policy, was silent on the NSB’s latest report, but in previous instances it has dismissed such claims as alarmist propaganda. Chinese officials insist that Taiwan is part of China and that all activities, including media reporting and social media activity, are intended to promote national reunification and counter efforts to split the country. From Beijing’s perspective, there is no “interference” in Taiwan’s internal affairs because Taiwan’s affairs are China’s affairs. That ontological gap — between viewing the democratic election as the internal process of a self-governing polity and viewing it as a separatist challenge to Chinese sovereignty — makes any meaningful exchange on the issue virtually impossible. The NSB report, therefore, is not just a catalog of malicious behavior; it is also an argument for vigilance, resilience, and international support, as Taiwan’s government faces constant pressure to defend its democratic institutions from technological and narrative attacks. It also reflects the island’s broader struggle to maintain its autonomy in a region where fact and fiction are increasingly weaponized.
One of the most alarming findings in the NSB report is the extent to which generative artificial intelligence has become a weapon in the campaign. The bureau told lawmakers that Beijing is not merely producing AI-generated videos but is actively using AI to gather and process information on Taiwanese current affairs, candidates’ policy positions, and public polling data. This intelligence feeds the design of influence operations, allowing propagandists to identify which issues resonate most strongly in each constituency and to modify their talking points in nearly real time. The report warns that AI could be used to create fake but credible video and audio of politicians making incendiary statements, fabricate online identities capable of passing automated verification, and generate massive volumes of personalized messages sent by email or direct message to individual voters. While the report does not name specific cases, it notes that the accelerated use of AI coincided with the surge in detected threats beginning in June, with the number of suspect items peaking in the months leading up to the formal start of the campaign period. The target set is broad: every voter who may be susceptible to an attack on a candidate, a misleading claim about a policy, or a divisive rumor about a social group. In the context of Taiwan’s local elections, where turnout is often decided by passion and frustration, even a small shift in perception in a tightly contested municipality can change the outcome. The “nine-in-one” elections — so named because nine types of local positions are chosen on the same day — have been a priority for Beijing for decades. They determine who controls the levers of local government, which in turn shape Taiwan’s economic relationship with the mainland, the management of cross-strait cultural exchanges, and the environment for investment. The NSB’s warning suggests that Beijing views these elections not merely as a chance to install friendly officials, but as an opportunity to reframe the island’s political conversation, normalize Chinese political expressions, and delegitimize the DPP’s “resistance” narrative. As the bureau put it, Beijing has “expanded the use of AI technology to manipulate information,” and Taiwan’s ability to detect and respond to such manipulation remains an arms race with an opponent that has seemingly endless resources.
The NSB’s findings echo a broader pattern documented by independent journalism. A Reuters investigation conducted with Taipei-based research group IORG found that Chinese state media have increasingly amplified Taiwanese critics of the ruling Democratic Progressive Party in an information war designed to magnify domestic dissent. The investigation identified influencers, commentators, and politicians linked to the opposition Kuomintang — Taiwan’s oldest political party and the one most associated with a policy of closer ties with Beijing — being given prominent space in Chinese media and on Chinese social media. Their statements are often edited to highlight anti-DPP rhetoric and framed as evidence that Taiwan’s people reject the government’s line on sovereignty. The amplification is not accidental; it follows a playbook of long-term cultivation of “friendly” voices on the island, providing them a global reach that they might not otherwise enjoy and seeking to shape their political fortunes. In addition to KMT-linked figures, the operation targets civil society groups, trade bodies, and cultural personalities willing to express support for Chinese “reunification.” By giving these voices disproportionate attention, Beijing hopes to create an illusion of a powerful pro-unification bloc within Taiwan, while simultaneously framing DPP supporters as a noisy minority. The NSB report tells a similar story, describing coordinated efforts to legitimize pro-Beijing narratives and delegitimize critical voices through mass reporting, algorithm manipulation, and comment flooding. The report says Chinese state media serve as the core of this system — as both originator and validator — but that the actual dissemination relies on an ecosystem of anonymous accounts, borrowed identities, and even hacked pages. The result, the bureau argues, is not only an influence on the election outcome but also a slow degradation of Taiwan’s ability to have a coherent national conversation. People are pushed into polarized camps, trust in civic institutions erodes, and the idea that all information is merely manipulative becomes a self-fulfilling prophecy. This is exactly what Beijing wants.
Alongside the disinformation push, Taiwan’s intelligence agencies have charted a parallel spike in cyberattacks directed at government infrastructure. According to data submitted to the Legislative Yuan, Chinese cyberattacks on Taiwanese government agencies averaged 2.58 million per day in the first nine months of 2026. Detected threats surged from June, reaching 185.6 million in July and 106.4 million in August, according to the National Security Bureau. The bureau said while some of these attacks are routine espionage and criminal interference, the temporal correlation with the campaign season raises serious concerns. “It cannot be ruled out that this was advance intelligence-gathering and harassment preparation aimed at interfering in Taiwan’s year-end ‘nine-in-one’ elections,” the report said. The cyber and information operations appear designed to work in tandem: services hacked or flooded can be disabled at critical moments, while the disinformation machine spreads confusion about the cause of the disruption. An ordinary voter might receive a false message about a power outage, view a fake video of a politician admitting corruption, and then encounter a cyberattack that prevents the election commission from answering questions. The cumulative effect is a fog of uncertainty that favors the party best positioned to offer certainties — in this case, the Chinese Communist Party and its allies on the island. For Taiwan, the response to this multi-layered challenge involves improving digital literacy, strengthening fact-checking networks, coordinating with foreign intelligence and social media platforms, and criminalizing certain forms of manipulation. The NSB report is not a declaration of victory but a plea for continued vigilance. It argues that Beijing’s operations are dynamic: as one channel is shut down, another opens; as one troll network is identified, a new one appears, often with better encryption and more sophisticated AI. The 2.5 million distorted, fabricated, or deceptive items that Taiwan identified in just nine months are probably only a fraction of the actual total. Yet by exposing the scale and mechanics of the campaign, the NSB hopes to make the threat visible and to prepare the island’s institutions and people for a struggle that will not end with the November elections. As the report itself warns, in an era of AI-driven information warfare, the first casualty is not always truth — sometimes it is the trust that makes democracy possible.