“Data Center Bandwagon” Campaign: US-targeted influence activity
OpenAI ● Covered by 2 sources
OpenAI banned a group of accounts tied to China that used ChatGPT to churn out posts trashing US data centers. It's a fake grassroots pile-on dressed up as real local outrage over AI infrastructure.
Based on reporting by OpenAI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI says it has shut down a cluster of accounts, likely originating from mainland China, that leaned on its own chatbot to mass-produce social media content attacking American data center projects. The company calls the effort the "Data Center Bandwagon" campaign, and the mechanics are pretty familiar by now: generate posts, comments, and short articles that sound like ordinary Americans venting about noise, water use, or property values near a proposed AI facility, then spread them across multiple platforms to make the backlash look bigger and more organic than it is.
What's notable isn't the message so much as the method. These operators didn't invent new talking points about AI infrastructure. They used ChatGPT to draft, translate, and localize existing criticism at scale, then seeded it into US-facing accounts and forums. OpenAI's threat intelligence team flagged the coordinated posting patterns, shared language across supposedly unrelated accounts, and the sheer volume of near-identical framing showing up in different corners of the internet within short windows of each other.
This is the fourth or fifth such takedown OpenAI has disclosed this year involving suspected state-linked influence work, and the pattern keeps repeating: someone abroad uses a widely available US chatbot to sound more American than an actual American op would manage on its own. The irony is heavy. A campaign designed to stoke skepticism about US AI buildout was itself powered by US AI infrastructure, running on the same servers it was arguing against.
OpenAI didn't attribute the operation to a specific government agency, sticking with the more cautious "likely PRC-origin" language it tends to use when attribution is inferred from behavior and infrastructure rather than confirmed through other channels. The company says it banned the associated accounts and is sharing indicators with industry peers, which is standard practice now for any lab that catches this kind of activity — publish enough detail to help others spot the next version, without tipping off exactly how the detection worked.
My take — AI-written commentary, not fact-checked reporting
This is going to keep happening because the barrier to entry for astroturfing has basically collapsed — you don't need a troll farm anymore, you need an API key and a decent prompt. What worries me more than the China angle is that data center backlash is a legitimate, homegrown issue in plenty of US towns, and this kind of fake amplification makes it harder to tell real local anger from manufactured noise, which is exactly the point.
Read more about this at: OpenAI
