China was likely behind an online influence operation to sway U.S. perceptions of artificial intelligence technology and reshape the debate in Washington around the infrastructure needed to support it, according to research from OpenAI published Wednesday.
OpenAI said it caught the influence campaign because China-backed operatives were using ChatGPT to create content for the social media campaign.
The report’s findings are likely to further fuel claims made by Republicans and others pro-AI voices that foreign entities have tried to meddle in the debate over AI data centers, which has become a policy flashpoint heading into the midterms.
Still, OpenAI said it has not seen evidence that the campaigns succeeded in influencing public opinion on a widespread scale, according to Ben Nimmo, principal investigator of intelligence and investigations at OpenAI.
“Neither campaign appears to have gained much authentic engagement,” Nimmo said during a Wednesday press briefing. “They’re important for what they reveal about the intentions of influence operators from China, and the narratives they’re testing and seeking to amplify, but not for the impact.”
OpenAI’s researchers identified two clusters of ChatGPT users “likely originating from China” who used the AI chatbot to generate social media content “in support of apparent covert influence operations” promoting certain narratives about AI. This includes claims that data center build-outs are raising electricity costs for the average American family and that President Donald Trump has weaponized tariffs to keep the U.S. ahead in the global tech race.
These accounts have since been banned, the report said.
One cluster of users asked ChatGPT to generate images and comments pushing these narratives. These comments were then posted on social media by “batches of accounts” posing as Americans, Nimmo said.
Another cluster identified by researchers used AI to generate social media content criticizing the Trump administration’s tariffs as an attempt to “dominate technological competition.” Prompts used for this campaign were submitted in Simplified Chinese and asked that AI-generated content not include Chinese President Xi Jinping and focus solely on Trump — a possible tell that China was behind the operation, according to the report.
Nimmo said that the influence campaign amplified existing public backlash in the U.S. against the creation of new AI data centers, which has resulted in dozens of proposed moratoriums at the local, state and national level.
Policymakers have been fielding widespread constituent concerns about the environmental impacts and the added strain on local energy grids and water supplies brought about by power-hungry data centers.
”There is a genuine debate going on in the states over the future of AI and data centers … but what we don’t want to see is a covert foreign influence interference operation using the very same AI that it attacks,” said Nimmo.
A spokesperson for the Chinese embassy did not immediately respond to a request for comment.
The report comes a week after House Republicans urged the Trump administration to probe China’s role in allegedly stoking public opposition to AI data centers. They cited two separate recent reports from think tank Bitcoin Policy Institute and the right-leaning group Power the Future alleging that groups with foreign ties have played a role in driving anti-data center sentiments.
Sen. Dave McCormick (R-Pa.) echoed those claims Wednesday at POLITICO’s Energy Summit.
“There’s a reason that you’re seeing this incredible campaign of misinformation and all these things about data centers,” he told POLITICO’s Jonathan Martin. “It’s because the Chinese are behind a lot of it.”
House Energy and Commerce Chair Brett Guthrie (R-Ky.) also said Wednesday he intends to push ahead with legislation to address the impact of data centers as lawmakers consider ways to meet the massive amounts of energy required to power them.
EPA administrator Lee Zeldin, meanwhile, has endorsed letting state and local communities decide how to handle the issue and pledged that the Trump administration will not set nationwide environmental requirements or recommendations for data centers.
OpenAI’s findings are likely to bolster claims from some industry executives that foreign actors are working to stoke opposition to the energy and computing infrastructure underpinning America’s AI ambitions.
Kevin O’Leary, whose firm is developing a large AI-focused data center project in Utah, said he began seeing what he described as a coordinated wave of misinformation after the project was announced in May. The posts, he said, falsely claimed the facility would drain water from the Great Salt Lake, consume massive amounts of power and sprawl far beyond its actual footprint.
“I’ve seen everything in business, but I’ve never seen this,” O’Leary, chair of O’Leary Ventures, told POLITICO of the online backlash targeting his project. “We didn’t know who these people were, so we hired data scrapers, we looked at the IP addresses — and we started understanding that something nefarious was going on here.”
O’Leary said the efforts underscore how AI infrastructure has become a new front in the U.S.-China competition for technological and economic dominance.
“The country with the best AI will win the future wars and will have the most efficient and productive economy,” he said. “It is the underpinning of economic success, and the Chinese know that — we can’t let them win.”