OpenAI report links ChatGPT campaigns to China data-center push

OpenAI report – OpenAI says accounts likely tied to China used ChatGPT to draft English talking points and images for social media influence campaigns aimed at Americans. One cluster pushed “Data Center Bandwagon” messaging about electricity demand and higher bills, while ano

On a public timeline, the message looked familiar: energy costs, power demand, and the familiar anxiety that AI’s expansion will hit households where it hurts. OpenAI now says that familiar story line wasn’t just organic debate.

In a report shared by the company. OpenAI describes China-linked influence activity that it says used ChatGPT to plan and generate content designed to sway American opinions about AI data centers. The accounts it identified were divided into two clusters—one leaning into data center “electricity bill” arguments. the other attacking US tariffs and technology policy.

The first cluster OpenAI designated “Data Center Bandwagon.” Accounts in this group allegedly asked ChatGPT to produce English-language talking points and images. including comic strips. The material centered on a claim that AI data centers increase demand for electricity. which in turn leads to higher bills for consumers.

OpenAI says these users posed as Americans from a variety of backgrounds on social media. They posted the text and image outputs they generated with ChatGPT. The company believes they were part of a social media team at a private Chinese company working for local government clients.

The reach—and the intent—also extended beyond simply posting. OpenAI says the accounts uploaded a file to the chatbot outlining their objectives and strategies: how to sway public opinion and how to establish fake social media accounts without getting detected.

In the same report, OpenAI says the campaign targeted Chinese people based in other countries. It even claims they asked ChatGPT to generate insults meant to harass Chinese dissidents and political commentators. Still, the operation wasn’t limited to foreign-language aggression. When posing as US-based Chinese immigrants and professionals. the accounts also reportedly egged online public personalities to talk more about US policy failures.

What made the campaign easier to blend in. OpenAI argues. is that it used topics that people are already actively discussing. Even while OpenAI says the accounts were likely inauthentic. it says they posted links to “legitimate news stories about the power grid operator’s capacity auctions and data center power demand.” The company’s emphasis lands on a real-world issue: the rising cost of electricity in cities and towns near AI data centers.

A Bloomberg report mentioned in OpenAI’s coverage frames that risk in stark numbers: electricity can cost as much as 267 percent more for a month compared to five years ago in areas close to data centers, because demand outstrips supply.

OpenAI’s second cluster looks different, but the shape of the operation is familiar. The accounts it identified generated comments and images criticizing US tariffs and tech policies. The messaging, OpenAI says, pushed the idea that the United States has been “backstabbing allies.”

This group, too, asked ChatGPT to tailor content to particular audiences. OpenAI reports that they instructed the chatbot to keep Chinese President Xi Jinping out of the images they generated. They also asked for comments in English, Italian, Japanese, and traditional Chinese, aiming at Taiwanese audiences.

OpenAI also acknowledges that the campaigns didn’t succeed in changing minds at scale. The company says the efforts failed to gain much authentic engagement and haven’t exactly shifted public opinion.

The company’s central point is sharper than that. OpenAI says the significance lies in how the operators tried to hide their identities while inserting themselves into an ongoing American debate about the future of the country’s AI capabilities.

They reportedly attempted to do it without showing who they were or what motivated them—using an American chatbot to draft content, post through accounts meant to look real, and steer attention toward controversies that already had public traction.

And one detail remains oddly unanswered even inside OpenAI’s own report: OpenAI says it is not in a position to determine what drove the operators’ choice to use an American AI chatbot instead of another system, such as DeepSeek.

What OpenAI can say. based on its report. is that the campaigns were designed to look like conversation—not covert engineering. Yet the tactic was covert engineering all the same: using generated talking points. images. and multilingual comments to nudge the public story around data center power demand. electricity bills. and broader US policy fights—while allegedly trying to avoid detection.

OpenAI ChatGPT influence campaign China-linked AI data centers electricity costs power demand social media cybersecurity misinformation DeepSeek Xi Jinping tariffs tech policy