TAIPEI (Taiwan News) — OpenAI revealed that an individual allegedly associated with Chinese law enforcement used its ChatGPT in ways that exposed a large‑scale influence campaign in its latest report on AI misuse.
The campaign combined AI tools with networks of fake accounts and coordinated tactics to manipulate public opinion and silence critics online and offline. The company said the activity involved a “well‑resourced, meticulously orchestrated strategy for covert information operations” against domestic and foreign targets, involving at least hundreds of staff and thousands of fake accounts across dozens of platforms.
The operation targeted overseas Chinese dissidents and Japanese Prime Minister Takaichi Sanae. Activities included drafting status updates, editing reports, and coordinating online influence efforts, all designed to suppress dissent and shape global perception.
The report highlighted that the alleged Chinese operator repeatedly stressed the need to combine online and offline tactics. The user described the arrest of a young woman in China for posting a pro-Taiwan tweet and the distribution of hostile flyers near the homes of individuals critical of the Chinese government, which were then circulated online to appear authentic.
“This is what Chinese modern transnational repression looks like,” Ben Nimmo, principal investigator at OpenAI, told reporters. “It’s not just digital. It’s not just about trolling. It’s industrialized. It’s about trying to hit critics of the CCP with everything, everywhere, all at once,” CNN reported.
Targeting Takaichi and Taiwan-related comments
In mid-October, shortly before Takaichi became Japan’s first female prime minister, the individual asked ChatGPT to help plan a campaign to discredit her. OpenAI said the AI refused the request.
This followed Takaichi's public criticism of human rights conditions in Inner Mongolia, an autonomous region in northern China where authorities have faced international scrutiny for suppressing the cultural and educational rights of ethnic Mongols, according to the BBC.
OpenAI said the account sought help in drafting a plan centered on six main elements. The plan aimed to portray Takaichi as far-right, amplify negative commentary online, stir social media outrage, boost the visibility of attacks, and spread positive commentary about conditions in Inner Mongolia.
The campaign extended to her Taiwan remarks, with anti-Takaichi memes circulating online. She had said that Chinese military action against Taiwan could be considered a “survival-threatening situation” for Japan, potentially prompting the country to take military action.
In November, OpenAI identified accounts posting similar hashtags and memes, including English-language criticisms. The patterns suggested coordinated updates by the same operators, even though the posts themselves were not generated by ChatGPT.
Harassment of overseas dissidents
OpenAI’s report also described attempts to intimidate Chinese dissidents abroad.
On one occasion, operators posed as US immigration officials to warn a dissident in the US that their statements had violated the law. On another, US county court documents were forged to attempt social media takedowns, though the effort was never fully carried out.
The report detailed a much wider range of cyber special operations. At different times, the user referenced over 100 tactics designed to identify, pressure, disrupt, and silence critics.
These included manipulating narratives, amplifying or suppressing content, attacking dissidents’ legitimacy, and exerting social or psychological pressure. Some tactics targeted mental health, families, or social media accounts, while others involved hacking livestreams, creating websites outside China, or attempting to infiltrate Western platforms.
Operations spanned Chinese domestic networks such as Weibo and WeChat and more than 300 foreign platforms, generating millions of posts with hundreds of operators. AI tools were reportedly used for monitoring, profiling, translation, content creation, and internal documentation.
Michael Horowitz, former Pentagon official and current University of Pennsylvania professor, told CNN that OpenAI's report “clearly demonstrates how China is actively employing AI tools to enhance information operations. US-China AI competition is intensifying, not just at the frontier, but in the daily planning and execution of surveillance and information efforts.”
