OpenAI Uncovers China-Linked Campaign Targeting US AI Policy

OpenAI has identified what it describes as a China-linked influence operation that used its own ChatGPT platform to generate social media content aimed at shaping American perceptions of artificial intelligence infrastructure and policy.

The company disclosed Wednesday that it detected and banned two clusters of accounts originating from China that used the AI chatbot to create posts for what researchers characterized as covert influence campaigns. The content targeted debates around data center construction and the Trump administration's technology tariffs.

The mechanics of the operation

One group of users prompted ChatGPT to generate images and comments claiming that data center expansion is driving up electricity costs for American families. These AI-generated materials were then posted across social media platforms by accounts posing as US citizens, according to Ben Nimmo, principal investigator of intelligence and investigations at OpenAI.

A second cluster used the platform to produce content criticizing the administration's tariff policies as an attempt to dominate global technology competition. Notably, these prompts were submitted in Simplified Chinese and specifically requested that generated content avoid mentioning Chinese President Xi Jinping while focusing exclusively on President Trump—a detail researchers cited as evidence of Chinese state involvement.

Why it matters

The disclosure arrives as AI infrastructure has emerged as a contentious policy issue heading into midterm elections, with dozens of proposed moratoriums on data center construction at local, state, and national levels. OpenAI's findings provide concrete evidence that foreign actors view this domestic debate as an opportunity for strategic interference, even as genuine constituent concerns about environmental impacts and energy grid strain remain legitimate. The revelation that adversaries are weaponizing AI tools to influence AI policy debates adds a new dimension to discussions about both infrastructure development and platform security.

Limited impact, significant intent

Despite identifying the campaigns, OpenAI emphasized that neither operation appears to have achieved widespread influence on public opinion. "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," Nimmo said during a press briefing.

The company noted that the influence effort amplified existing public backlash against data center construction, which has generated legitimate constituent concerns about environmental impacts and strain on local energy grids and water supplies.

Political response

The report follows a letter from House Republicans last week urging the Trump administration to investigate China's alleged role in stoking opposition to AI data centers. House Energy and Commerce Chair Brett Guthrie said Wednesday he plans to advance legislation addressing data center impacts as lawmakers consider how to meet the massive energy demands of AI infrastructure.

Kevin O'Leary, whose firm is developing an AI-focused data center project in Utah, told the publication he observed what he described as coordinated misinformation after announcing the project in May. He characterized AI infrastructure as a new front in US-China technological competition, stating that "the country with the best AI will win the future wars and will have the most efficient and productive economy."

The Chinese embassy did not immediately respond to requests for comment on the allegations.

These details were first reported by Politico.