OpenAI says China-linked accounts stoked US data-centre backlash over electricity prices

OpenAI says China-based actors likely used ChatGPT to run covert influence operations aimed at turning Americans against AI data centres. In a research report released on Wednesday, the ChatGPT maker said it banned a cluster of accounts likely based in China for attempting to “manipulate a legitimate debate about American AI,” including content that blamed data centres for higher electricity bills.

The tactics were designed for social media speed and persuasion: OpenAI said the accounts generated social media comments and images across the United States, including a comic strip showing a cigar-chomping businessman holding bags marked with dollar signs as a family reacts in shock to their electricity bill, according to the San Francisco-based company. It also said a second cluster of accounts produced material casting US tariffs as an effort to “dominate technological competition” with China, with instructions that the content should not mention Chinese leader Xi Jinping.

This is the uncomfortable intersection of AI, infrastructure, and information operations. Data centres are the physical engine behind AI models like ChatGPT, and the political fight around them has been heating up in the US. According to Data Center Watch, a research project by AI security company 10a Labs, at least 36 projects were blocked or delayed between May 2024 and June 2025. That kind of friction creates real leverage for anyone trying to shift public sentiment, slow permitting, or inflame community opposition.

OpenAI’s report, however, draws an important line that decision-makers will care about: while it found the operations, it said it found no evidence the effort had a “meaningful” influence. The company framed the playbook in familiar terms to researchers of foreign interference. OpenAI said foreign influence operations often “latch onto existing local issues and sincerely held beliefs,” using credibility to amplify divisions or worsen public distrust. In this case, OpenAI said the operators tried to covertly insert themselves into an ongoing American debate about the future of US AI capabilities while hiding who they were and what motivated them.

That emphasis on “meaningful” impact is where boards and executives should pay attention. If public opinion is already primed, even small information campaigns can add noise. But OpenAI is telling the market something more subtle: detected activity does not automatically equal strategic control. Darren Linvill, a professor at Clemson University who studies foreign influence campaigns, told Al Jazeera he doubts that the campaign identified by OpenAI, or any similar coordinated effort, would have much impact on the “volume or tone” of the debate. Linvill said his team is familiar with Chinese influence actors and that the AI work China has done “has been interesting but not effective,” adding that it is improving but “they aren’t there yet.” He also questioned the logic of using OpenAI itself if the goal were meaningful influence, saying, “If China were really serious about meaningfully influencing the discourse around data centres using AI chat bots, I question if they would use OpenAI to do it.”

OpenAI is also not the first institution to put foreign influence on the table. In May, Secretary of the Interior Doug Burgum told a policy event hosted by Breitbart News that negative sentiment toward data-centre construction is not “organic” and could, in some cases, be linked to “foreign-sourced dark money.” And this matters for executives because the claim, even if disputed, can shape how regulators, lawmakers, and the public interpret any protest, delay, or policy dispute. In other words, the information environment becomes part of the operating environment.

The policy backdrop adds further urgency. Opposition to data centres is tied in part to energy demand. The facilities accounted for 1.5 percent of global electricity use in 2024, with consumption growing 12 percent annually over the last five years, according to the International Energy Agency. That is a measurable stress point that can be debated on facts, not just frames. Yet politics rarely sticks to spreadsheets. In March, Senator Bernie Sanders and House Representative Alexandria Ocasio-Cortez announced legislation that would impose a moratorium on new data centres until national safeguards exist to mitigate AI risks. The bill’s path is described as unlikely in the near term due to President Donald Trump’s laissez-faire approach to AI regulation and Republican control of both chambers of Congress.

Now add the OpenAI angle. If AI models can be used for covert influence operations, then the compliance question for the industry is no longer only about model safety in a technical sense. It is also about how content is generated, amplified, and used in real-world disputes over power, jobs, and community impact. Even if OpenAI says it found no meaningful influence this time, executives still have to plan for the second order problem: faster campaigns, more believable targeting, and more sophisticated attempts to exploit public concerns.

For leaders building or funding AI infrastructure, the strategic stake is straightforward. Permitting and community acceptance can determine timelines as surely as supply chains. Whether or not the specific operation OpenAI described moved the needle “meaningfully,” it underscores a world where AI is not just a product. It is an instrument that can be repurposed in the information battlefield around critical infrastructure.