X says its Safety team has dismantled part of a massive bot farm: a network of 200,000 suspected inauthentic Chinese accounts, 200 of which were posting comic strips, images and text about AI data centers in a way that, per the company, "could manipulate a legitimate debate about American AI and energy policy."
The findings, announced on X's Global Government Affairs page and reported by Engadget, are the latest evidence that the fight over data center construction has spilled into coordinated influence operations. The suspicious posts argued that AI data centers drive up household electricity prices and strain the grids of nearby towns. They also included AI-generated cartoons depicting data center operators as unscrupulous businessmen raking in profits while average families shoulder the bill. For more on how AI is colliding with the energy economy, follow our AI news updates.
ChatGPT Was Reportedly the Propaganda Tool
Based on the sample posts X shared, the inauthentic accounts used OpenAI's tools to generate illustrations and talking points for their campaign. The same illustrations appeared in a report OpenAI published in June about accounts likely based in China using ChatGPT to generate anti-data center content.
That June report contained a nuance worth remembering: along with the generated propaganda, the accounts also shared "legitimate news stories about the power grid operator's capacity auctions and data center power demand." Influence operations work best when they are padded with real information — and in this case, the underlying information is genuinely contested. OpenAI noted at the time that while the ChatGPT users likely relied on fake accounts to spread their material on social media, they grounded their messaging in authentic coverage of grid capacity auctions and power demand, giving the campaign a veneer of credibility that pure fabrication would lack.
From June Report to August Rollup
Read together, the two disclosures sketch a campaign that has now run for at least a quarter. In June, OpenAI documented the production side: accounts likely based in China using its models to draft and illustrate anti-data center content. In August, X documented the distribution side: a slice of a 200,000-account network amplifying that same material, complete with the same illustrations OpenAI had already flagged. The through-line — generative AI as the content engine for influence operations — is what makes this rollup more than a routine platform moderation note.
The Backdrop: A Real Energy Debate
There are legitimate, well-documented concerns about power costs near hyperscale data centers. According to Goldman Sachs, electricity prices had already risen 6.9 percent year-over-year as of February 2026, and the bank expects prices to keep climbing as AI facilities add demand. Communities from Virginia to Arizona have organized against new campuses, and utilities have revised forecasts to accommodate the load.
That reality is what makes the X investigation awkward for everyone involved. The platform's announcement lands in a debate where critics of data center buildout have substantive grievances — meaning both the existence of the bot farm and X's framing of it will be contested. X did not list the accounts involved, though it posted examples of the content.
"We remain committed to maintaining an open and authentic platform where people debate topics of public interest," the company wrote, adding that it suspends accounts that violate its Authenticity policy.
What X Did Not Say
Significant questions remain unanswered. X did not explain what the other 199,800 accounts in the network were posting about, leaving open whether the data center content was the farm's primary mission or one project among many. The company also did not detail its attribution methodology or say how many accounts have been suspended.
The gap between the headline number and the confirmed activity — 200 of 200,000 — is itself notable. Influence operation rollups often work this way: the full network is mapped, but only the subset tied to a specific campaign is publicly described.
Why It Matters
The episode shows how the American AI debate is now a target for foreign information operations from at least two directions at once. OpenAI's June report documented China-linked accounts amplifying anti-data center narratives; the strategic reading is that slowing US data center construction serves competitors who face fewer constraints on their own buildouts. Whether or not that motive is correct, the practical effect is the same: authentic local concerns about power bills get mixed with manufactured outrage, making it harder for policymakers to read public sentiment.
For now, the verified facts are narrow: a platform identified an inauthentic network, attributed it to suspected Chinese actors, and linked its creative assets to an AI vendor's tooling. X has suspended accounts that violated its Authenticity policy, but the economics of bot farms favor persistence over any single takedown — rebuilding a network is cheaper than investigating one. The broader lesson — that generative AI has industrialized the production of political cartoons, talking points and plausible news framing — applies well beyond this single takedown, and it guarantees that both the content of the AI debate and the provenance of that content will remain contested terrain.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
