OpenAI has published a report detailing how users it believes were based in China used ChatGPT to generate content aimed at swaying American public opinion against AI data centers and US tariff policies. The company divided the accounts into two clusters and banned them, marking the first known instance of OpenAI's models being used in a Chinese foreign influence campaign, according to a spokesperson who spoke to Axios, as reported by TechRadar.
The Data Center Bandwagon campaign
The first cluster, designated the "Data Center Bandwagon" group, asked ChatGPT to generate English-language talking points and images, including comic strips, that focused on how AI data centers drive up electricity demand and lead to higher consumer bills, according to Engadget. The users posed as Americans from various backgrounds on social media and posted the AI-generated text and images. OpenAI believes these accounts were operated by a social media team at a private Chinese company working for local government clients. The company noted that the users even uploaded a file to ChatGPT describing their objectives and strategies for swaying public opinion and evading detection.
This group also targeted Chinese people living abroad, generating insults meant to harass Chinese dissidents and political commentators, Engadget reported. While posing as US-based Chinese immigrants and professionals, they encouraged online personalities to speak more about US policy failures.
Despite the effort, the campaigns failed to gain much authentic engagement and did not shift public opinion significantly, OpenAI acknowledged in its report, according to both Engadget and TechRadar. The content focused on real issues that were already controversial, such as the rising cost of electricity near data centers. According to a Bloomberg report cited by Engadget, electricity now costs as much as 267% more for a month compared to five years ago in areas close to data centers, because their energy demands outstrip supply.
The Tech and Tariffs campaign
The second cluster of accounts generated comments and images criticizing US tariffs and tech policies, TechRadar reported. This group specifically focused on content critical of the US and emphasized that the country has been "backstabbing allies." The users asked ChatGPT to keep Chinese President Xi Jinping out of the images they generated and instructed the chatbot to write comments in English, Italian, Japanese, and traditional Chinese to target Taiwanese audiences, according to Engadget.
OpenAI named this campaign "Tech and Tariffs," and it aimed to exacerbate existing tensions around trade policy, according to TechRadar. The company said these accounts were also part of a Chinese influence operation.
| Campaign | Goal | Content Type | Target Audience | Status |
|---|---|---|---|---|
| Data Center Bandwagon | Oppose AI data centers, highlight energy costs | Talking points, comic strips, insults | US public, Chinese dissidents | Banned, low engagement |
| Tech and Tariffs | Criticize US tariffs, tech policy | Comments, images | US, Italian, Japanese, Taiwanese audiences | Banned, low engagement |
Republican response and real concerns
The OpenAI report adds some validity to recent Republican claims that the anti-data-center movement is being fueled by China, but it does little to address the tangible effects of data center projects on local communities, TechRadar noted. A group of Republicans recently called upon FBI Director Kash Patel to investigate anti-data-center sentiment, alleging that the rising opposition is fueled by China, citing "language too similar to be coincidental" around water usage, energy constraints, transparency, and utility bills.
The negative effects of data center construction and the additional costs imposed by tariffs are existing areas of contention, not narratives invented by China, according to OpenAI. The campaigns were designed to increase the scale and visibility of these issues on social media platforms such as X, TechRadar reported.
Implications for enterprise security
For enterprise technology leaders, this incident underscores the growing sophistication of AI-powered disinformation campaigns and their potential to manipulate public opinion on topics that affect business operations, such as data center siting and energy costs. While these particular campaigns failed, the use of large language models to generate convincing, locally tailored content at scale presents a new vector for influence operations. Organizations should be aware that AI-generated disinformation could target supply chain infrastructure, trade policy debates, or regulatory environments. OpenAI's detection and ban of these accounts shows that AI companies are monitoring misuse, but the ease with which ChatGPT can generate such content highlights the need for robust monitoring and verification tools in enterprise contexts.
Sources:
