OpenAI says a group of China-linked operators attempted to exploit growing public concerns about electricity costs and environmental impacts by using ChatGPT to generate content opposing AI data centers in the United States. The campaign appears to have had little impact, but it offers a revealing glimpse into how artificial intelligence is becoming a tool in global influence operations.
OpenAI has disclosed that it disrupted several accounts linked to suspected Chinese influence operations that used ChatGPT to generate content aimed at shaping public debate in the United States.
According to a report released by the company, the activity focused heavily on controversial issues already dividing American communities, particularly the rapid expansion of AI data centers and the growing demand for electricity needed to power them.
The company said one of the campaigns, internally labeled “Data Center Bandwagon,” involved users generating English language social media posts, comments, and images that criticized the construction of AI infrastructure across the United States.
Many of the messages argued that data centers would drive up electricity prices, place additional strain on power grids, and harm local communities. OpenAI believes the operators then attempted to distribute the content online while posing as ordinary Americans.
The findings arrive at a time when the United States is investing heavily in AI infrastructure.
Major technology companies are racing to build new facilities capable of supporting increasingly powerful AI models. Those projects have sparked legitimate concerns among residents and environmental groups about water consumption, energy demand, land use, and rising utility costs.
OpenAI says the suspected influence operation sought to amplify those existing concerns rather than invent entirely new ones.
The campaign did not create the debate. According to OpenAI, it attempted to exploit a real debate that was already taking place across the United States.
The report identified another China-linked effort that focused on trade policy and tariffs.
In that operation, users allegedly employed ChatGPT to generate slogans, multilingual comments, and political cartoons criticizing U.S. trade policies and tariffs imposed by President Donald Trump. Some of the content was later shared on social media platforms including X.
One recurring theme involved portraying Trump’s economic policies as harmful to America’s global standing. OpenAI said the campaign produced content in multiple languages, including Chinese, Japanese, Italian, and English.
Despite the sophistication of the effort, OpenAI stressed that neither campaign appears to have achieved significant reach.
Investigators found little evidence that the generated content gained meaningful traction beyond the accounts involved in distributing it. The company described the operations as relatively small and largely ineffective.
That has not reduced concerns among security experts.
The report highlights how foreign actors are increasingly experimenting with generative AI tools to influence political discussions, shape public opinion, and amplify existing social divisions. Rather than creating entirely fabricated narratives, many campaigns now focus on taking genuine concerns and pushing them into more polarized territory.
The Chinese Embassy in Washington rejected suggestions of wrongdoing, saying it opposes the spread of false information and supports the responsible development of artificial intelligence.
For OpenAI, however, the discovery serves as another warning about how quickly AI technology is becoming embedded in geopolitical competition.
What makes the case unusual is the irony at its center: suspected foreign operators allegedly used an American AI system to generate content aimed at influencing American public opinion on the very infrastructure needed to power the future of artificial intelligence.
As governments and technology companies continue investing billions into AI development, experts expect similar influence campaigns to become more common.
The operations uncovered by OpenAI may not have changed public opinion in any measurable way, but they offer a glimpse of how future information battles could be fought not just through human propaganda, but through AI-generated content designed to blend seamlessly into everyday online conversations.
