Two hundred accounts out of 200,000. That’s the slice of a suspected Chinese bot network that X’s Safety team says was actively pushing anti-AI data center content on the platform. The rest of those accounts remain unexplained, which raises more questions than X’s report actually answers.
As reported by Engadget, X identified the network through an investigation into suspected inauthentic Chinese accounts and posted its findings on its Global Government Affairs page. The 200 active accounts were sharing comic strips, AI-generated cartoons, and written posts framing data centers as a threat to ordinary Americans, specifically around electricity costs and grid strain in communities near large AI facilities. The cartoons depicted data center operators as profiteering businessmen while average families foot the bill.
What makes this operation harder to dismiss than a typical influence campaign is that the underlying concerns are real. Goldman Sachs data puts year-over-year electricity price increases at 6.9 percent as of February 2026, and demand from hyperscale AI facilities is widely expected to keep pushing that number higher. Whoever ran these accounts didn’t fabricate a problem out of thin air. They took a legitimate policy debate and tried to shape it through fake personas, which is a much more sophisticated play than pure disinformation.
The ChatGPT connection is notable. X’s sample posts matched illustrations that OpenAI itself flagged in a June report about China-linked accounts using the chatbot to generate anti-data center content. OpenAI noted at the time that the same users also asked ChatGPT to write insults targeting Chinese dissidents and criticisms of US tariffs and tech policies. So this wasn’t a single-issue operation. The data center angle was one thread in a broader influence effort.
For the AI industry, the implications run in two directions. First, this is a reminder that AI tools are as useful to influence operators as they are to legitimate users. Generating consistent visual content across hundreds of accounts, at scale, used to require real resources. Now it’s a ChatGPT prompt. Second, the specific target matters. AI infrastructure and energy policy is a pressure point right now, with real regulatory and public relations exposure for companies building or expanding data centers across the US.
X said it suspends accounts that violate its Authenticity policy but declined to publish the full list of accounts involved. That’s a familiar move, and it limits outside researchers from verifying the scope or methodology of the investigation. Still, the overlap with OpenAI’s earlier findings gives this report more credibility than a self-published platform claim would normally carry on its own.
