OpenAI said on the 10th (local time) that it had detected signs that China-linked actors abused ChatGPT to conduct an influence operation in the United States.
In its "June 2026 threat report" published that day, OpenAI said influence operators backed by China tried to manipulate public opinion related to U.S. President Donald Trump's tariff policy and the construction of U.S. artificial intelligence (AI) data centers. According to the report, from late last year to early this year they used ChatGPT to produce social media (SNS) posts and images criticizing the Trump administration's policies and posted them on X (formerly Twitter).
One group produced content claiming that AI data centers raise electricity bills for U.S. households and cause harm, leading efforts to build opposition to data center construction. In the United States, more than 10 states are pushing or considering regulations over data center construction due to power consumption and environmental impact.
Another group tried to sow division by creating cartoon images asserting that President Trump is weaponizing tariffs to secure an edge in the U.S.-China tech hegemony race. According to the report, the group mass-generated political satire cartoons on the U.S.-China tech competition using ChatGPT, entering a command that said, "Only President Donald Trump should appear in the cartoon, and Chinese President Xi Jinping must not appear." The cartoons showed President Trump smashing a wall labeled "Global Future" with a hammer or sawing off the ladder he was standing on.
According to the report, they used virtual private networks (VPNs) to disguise their access as coming from outside China, but the prompts they entered into ChatGPT were written in simplified Chinese. They referred to themselves as "shuijun" ("water army"), which it added is Chinese slang for online accounts that carry out organized influence operations. The report assessed that they were likely affiliated with private corporations working on commission from Chinese local governments.
OpenAI said it had blocked all related accounts.
OpenAI assessed that these influence efforts had no real impact, but said the case shows how Generative AI is being actively used as a propaganda tool for influence operations. Ben Nimmo, OpenAI's lead Researcher, said, "It is ironic that Chinese actors tried to use U.S. AI technology to interfere in U.S. public opinion."
Regarding this, the Chinese Embassy in the United States countered that it "firmly opposes groundless attacks or accusations against China," Reuters reported.