By AJ Vicens, Hina Suzuki and Kentaro Sugiyama
Feb 26 (Reuters) - In the days surrounding Japanese Prime Minister Sanae Takaichi's February election win, several dozen X accounts linked to a Chinese misinformation campaign attacked her deeply conservative views and hawkish approach to China, said a U.S. research institute focused on national security and foreign policy.
The 35 accounts, along with nine channels on the microblogging site Tumblr, pushed corruption allegations and portrayed Takaichi as illegitimate and militaristic. The accounts suggest the prime minister is a reckless "cult-backed" leader driving Japan toward war, said Maria Riofrio, a researcher with the Foundation for Defense of Democracy's Center on Cyber and Technology Innovation.
The accounts, part of a network of at least 327 X and other social media accounts, have since December or earlier attacked adversaries of Beijing or pushed pro-China policy positions, and targeted human rights organizations and sought to influence domestic politics in Japan, the United States, the Philippines and Latin America, according to a new analysis by Riofrio.
A spokesperson for Takaichi said her office is aware of suspicious foreign social media accounts that have posted content related to Japan's elections.
"We consider this to be a national security threat that undermines the very foundations of democracy, including the fairness of elections and freedom of the press. We believe that countermeasures must be urgently prioritized."
Spokesperson Liu Pengyu for the Chinese Embassy in Washington said the "analysis" by the FDD was "groundless."
"The Chinese government consistently opposes and combats the use of fake accounts and other tactics to manipulate public opinion or spread disinformation," Liu said in a statement to Reuters.
"We urge the relevant parties to stop making unfounded accusations and smearing others based on speculation."
Riofrio identified the most recent campaign as a distinct cluster based on its pro-China narratives that has coordinated on messaging content containing overlapping hashtags and other similarities. The cluster is likely part of long-running Chinese information operations that internet security analysts have dubbed Spamouflage or Dragonbridge, she said.
Although engagement with the material is low, users do see the content in social media feeds and the operators work algorithms to boost the content. One tweet that accused Takaichi of being in a cult, for example, received only two likes but was viewed more than 1,000 times.
The activity demonstrates "China has the political will to interfere in Japanese elections and internal affairs," Riofrio said, noting that the overall cluster has similar operations targeting other countries, including the United States.
The FDD is a Washington-based nonprofit focused on "strengthening U.S. national security and reducing or eliminating threats posed by adversaries and enemies of the United States and other free nations," according to its website.
ATTACKS ON TRUMP, ASIA-PACIFIC NATIONS
The Nikkei newspaper reported earlier this week on another set of suspected foreign-backed online operations attacking the Japanese elections.
Riofrio said nearly half of the 327 accounts attacked U.S. President Donald Trump, pushing the narrative that his drug and border policies have worsened America's fentanyl crisis, reversing gains made during the Biden era, while also deflecting blame from China, according to the FDD analysis.
In a coordinated sequence of messaging in early February on Trump and fentanyl, six accounts, despite having fewer than 10 followers each, attracted hundreds of likes, retweets and replies and nearly 18,000 views as of February 12, Riofrio said.
One account in the network, FentanylFreeA, created in December 2025, Riofrio said, seemingly seeks to emulate the U.S. Drug Enforcement Agency’s Fentanyl Free America campaign, using a similar name and identical imagery. The account attacks both the U.S. and India, which it blames as the source of fentanyl precursor drugs.
Neither the White House or the DEA responded to a request for comment.
The operations known as Spamouflage or Dragonbridge have been active since at least 2017 and have repeatedly targeted audiences around the world over the years, according to experts.
A spokesperson for Google’s Threat Intelligence Group said Dragonbridge is the most prolific pro-PRC information operations operator that it tracks as of early 2026 based on “its massive scale and assertive narrative agenda.”
Dragonbridge, while maintaining a foundational focus on targeting the U.S., overseas dissidents, government critics and international NGOs, has become notably more assertive in the Asia-Pacific, the Google spokesperson said, including by targeting the political leadership of Japan, Japan-Taiwan relations, Vietnam over its South China Sea activities, India and the Philippine administration.
OpenAI on Wednesday reported that it had disrupted attempts in mid-October by a Chinese law enforcement official to help plan a multi-stage information attack on Takaichi.
(Reporting by AJ Vicens in Detroit; Additional reporting by Kentaro Sugiyama and Hina Suzuki in Tokyo; Editing by Chris Sanders and Tom Hogue)
