A ChatGPT user with links to Chinese law enforcement tried to use the AI chatbot to run smear campaigns targeting the Japanese prime minister and other critics of the Chinese Communist Party, according to OpenAI’s latest report on malicious uses of its models.
The user, since banned by the AI giant, tried to convince the model to help them plan a smear campaign against Sanae Takaichi, Japan’s first female prime minister, after she criticized the Chinese Communist Party for trampling human rights in Inner Mongolia.
These are basically covert influence operations and transnational repression
This occurred in mid-October, we’re told. Some of the prompts included asking the model to post and amplify negative comments about her on social media, and use fake email accounts purporting to be foreign residents to send complaints about Takaichi to other Japanese politicians.
When ChatGPT refused, the user seemingly used other companies’ models to carry out the campaign. Later that month, however, the same user asked ChatGPT to edit status reports on what they called “cyber special operations.”
“Despite the name, these are basically operations designed to harass and silence domestic and foreign critics, closer to what we’d think of as covert influence operations and transnational repression,” Ben Nimmo, principal investigator on OpenAI’s Intelligence and Investigations team, said on a call with reporters.
The malicious activity OpenAI spotted included generating status reports on operations targeting Chinese dissidents and CCP critics, along with the specific covert op against Takaichi. The latter seemed to follow the structure of the original draft plan to discredit the Japanese politician, focusing on the same five areas – negative comments, immigration, living conditions, far-right links, and tariffs – and even provided the threat intel team with several operational details about the smear campaign.
The status reports claimed the operation asked unnamed Japanese influencers for support, and launched a set of hashtags, including #右翼 共生者 (this means “right-wing symbiont”). OpenAI says it found evidence of this hashtag spreading – but in “small quantities” – across X, Japanese online community Pixiv, and Blogspot beginning in late October 2025.
None of the posts containing the hashtag alongside Takaichi memes gained much traction. The YouTube videos had single-digit views, while Xeets and Pixiv posts typically showed zero engagements. One meme posted to Pixiv scored 108 views, the biggest audience for this hashtag.
From social media to psychological pressure
The same user’s broader activity across a wide range of “cyber special operations” moved beyond posting on social media and into exerting social and psychological pressure to silence critics. The attacker’s tactics included targeting dissidents’ mental health and their families, hacking their livestreams, and reporting their social media accounts for phony violations, sometimes supported by fake evidence.
In one such incident documented in the ChatGPT user’s diary of cyber ops, they created a fake obituary and gravestone photos claiming that dissident Jie Lijian had died before mass posting these messages online.
Another report detailed efforts to get activist Hui Bo (@huikezhen) removed from X by filing thousands of reports against his Xeets and creating dozens of fake accounts using his likeness.
“While we are not able to independently confirm whether and how any such abusive reports were actually sent, as of November 29, 2025, Hui’s X account was indeed restricted, and a number of other X accounts that used his name and profile picture showed up in search results instead,” OpenAI’s report notes.
One another occasion, the same user reported that public security services targeted multiple dissidents because of a single campaign in which operators created a series of fake claims accusing three dissidents of a sex scandal. Online searches for the three names together produced “multiple” pieces of content that mirrored this claim across various blogs, Reddit, YouTube, Tumblr, Adobe’s community for creators Behance, and other websites.
“This is what Chinese, modern, transnational repression looks like,” Nimmo said. “It’s not just digital. It’s not just about trolling. It’s about trying to hit critics of the CCP with everything everywhere, all at once.”
These cyber operations are “well -esourced” and “meticulously planned,” he added. “They target people who dare to criticize the CCP’s record, not just at home, but anywhere in the world.”
This activity echoes earlier China-based influence ops campaigns dubbed “Spamouflage” by research teams. That the user’s ChatGPT inputs included hashtags and references to fake social media accounts indicate “much wider cross-internet activity,” the report says.
Meta, in its August 2023 threat report [PDF], attributed Spamouflage to individuals connected to Chinese law enforcement. ®