OpenAI has identified coordinated attempts by Chinese-speaking actors to exploit its ChatGPT platform to generate material opposing U.S. trade and technology policy, the company said in a report published on Wednesday. According to OpenAI, the activity, which it dates to late 2025 and early 2026, focused on producing slogans, cartoons and multilingual comments designed to influence public debate on tariffs and the domestic siting and regulation of data centers.
The company said one cluster of users used ChatGPT to create visual and textual content that was later posted to X. Those cartoons depicted a disruptive political figure attacking international institutions and infrastructure - including images of the figure swinging a hammer at a barrier representing the future and sawing apart a ladder he was standing on. OpenAI said the same user group also produced Chinese-language comments for placement beneath Chinese-language articles, and generated material in Italian and Japanese.
OpenAI said it had subsequently traced another set of users to a Chinese technology company that does government work, though it did not identify the firm. That second group, the company said, attempted to shape U.S. debates over AI and data centers - a contentious domestic issue, OpenAI noted, where more than a dozen U.S. states have or are weighing restrictions on data center construction.
Ben Nimmo, a principal investigator at OpenAI, told journalists the operations appeared tailored to manipulate "a legitimate debate about American AI and wider American tech policies. Under the circumstances it’s particularly ironic that they tried to use American AI to do it." OpenAI said the influence operations showed little evidence of having achieved significant traction.
The sample imagery OpenAI shared painted the technology industry as profiteering and depicted data centers as imposing high electricity consumption that harmed ordinary citizens. OpenAI framed these creative outputs as examples of how AI-generated imagery has become pervasive, even when employed in material critical of the industry behind the technology.
OpenAI said it reached out for comment to relevant parties; the Chinese Embassy in Washington did not immediately return an email seeking comment, and X’s owner, xAI, also did not immediately respond to an email requesting comment.
The company highlighted that the use of its systems in these influence efforts - while apparently not successful - underscores the central role generative AI tools are beginning to play in digital influence campaigns. OpenAI’s findings also align with reporting that digital content farms have been using AI-generated imagery to produce memes and other material opposing data centers.
Context for industry observers
From an industrial perspective, the episodes reported by OpenAI touch on several nodes of the tech economy: public perception of data center energy use, policy disputes over infrastructure siting, and reputational pressure on AI companies. While OpenAI reports limited measurable impact from the operations it identified, the incidents illustrate a convergence between generative AI tooling and organized messaging efforts aimed at influencing domestic policy debates.