OpenAI says China-linked accounts used ChatGPT to seed a covert campaign about AI data centers
OpenAI has published what it describes as evidence of two China-linked ChatGPT account clusters that used the model to generate social media comments and images aimed at shaping U.S. debates over AI infrastructure, energy prices, tariffs, and broader technology policy.
The most concrete target was the buildout of AI data centers. OpenAI says one cluster, which it calls the “Data Center Bandwagon” campaign, pushed narratives claiming that AI data centers were raising electricity prices for ordinary families. The company says the accounts used ChatGPT to produce talking points, images, and posts that posed as Americans while criticizing the power demands of AI infrastructure.
The episode matters less because OpenAI says the campaigns achieved broad influence, and more because it shows how generative AI can be used to mass-produce plausible political content around a real public concern: whether the country’s AI boom is quietly shifting local costs onto households and communities.
What OpenAI says happened
OpenAI’s June 10, 2026 report describes two clusters of ChatGPT accounts it believes likely originated in China. The company says both were banned after using its models to support apparent covert influence operations.
The first cluster generated social media comments and images claiming that AI data center buildouts were increasing electricity prices for average families. OpenAI says these accounts posed as Americans from different backgrounds and posted text and image output they had generated through ChatGPT.
The second cluster focused on U.S. tariffs and tech policy, generating comments that criticized American leadership and emphasized claims that the United States had been “backstabbing” allies. OpenAI says this cluster also targeted Taiwanese audiences and instructed the model to avoid showing Chinese President Xi Jinping in generated images while including only U.S. President Donald Trump.
OpenAI says the operators also targeted OpenAI directly by spreading false claims that ChatGPT user data had been compromised. The company says those allegations were entirely false.
Why data centers became the target
AI data centers are one of the most visible front lines of the AI boom. They require enormous amounts of electricity, water, land, and grid capacity. As a result, communities near major deployments are already debating whether the benefits of AI infrastructure outweigh local costs.
That makes data centers an attractive target for influence operations. The concern does not need to be invented from whole cloth: energy prices, grid reliability, and local infrastructure strain are real issues. OpenAI’s report says the operators attempted to exploit existing concerns about energy prices and local impacts of data center development.
Engadget’s reporting adds one concrete example: it cites a Bloomberg report saying electricity costs in areas close to data centers can rise sharply compared with five years earlier because energy demand outstrips supply. Even if the exact causal chain is complex, that local pressure gives political narratives a ready-made hook.
What this reveals about AI-enabled influence
The report is a reminder that modern influence operations do not need to invent controversy from scratch. They can attach themselves to legitimate debates, then use AI tools to scale the volume and polish of messaging.
In this case, OpenAI says the operators used ChatGPT to generate English-language talking points and images. That matters because AI-generated content can lower the cost of producing persuasive material at scale, especially when paired with fake or semi-fake social media accounts.
The company also says the campaigns failed to gain much authentic engagement and did not shift public opinion. That is important context: this was not a viral takeover, but it still shows the playbook—use real anxieties, hide the sponsor, and flood a debate with AI-assisted content.
Bottom line
OpenAI’s report is not just about China, ChatGPT, or data centers individually. It is about the convergence of those things: a real infrastructure debate being targeted by AI-assisted influence operations. The practical takeaway is that the public should treat claims about energy prices, grid strain, and AI infrastructure with both skepticism and seriousness. The local costs are real enough to be politically potent; the covert campaign behind them is exactly why transparency matters.