Summary
OpenAI’s June 2026 threat report reveals that the company identified and disrupted a covert China-linked influence operation that used ChatGPT to generate anti-data center disinformation targeting American audiences. The campaign involved accounts believed to originate from a private Chinese technology company working for provincial government clients, using VPNs and Simplified Chinese prompts to generate English-language content.
The primary narrative pushed by the operation was that rapid AI data center construction in the United States was causing electricity prices to soar for ordinary families. The actors created fake social media accounts impersonating “ordinary Americans” on X (formerly Twitter) and Facebook, deploying AI-generated text, comments, and even comic strips to simulate grassroots opposition. The campaigns also extended to criticizing U.S. tariffs and running smear campaigns against Japanese politicians and Chinese dissidents.
OpenAI reports that the influence operations had minimal real-world impact, failing to generate significant engagement or shift public opinion. The company has banned all accounts involved. However, the report emphasizes that the campaigns demonstrate active foreign interest in testing disinformation narratives against critical U.S. AI infrastructure.
Source
OpenAI June 2026 Threat Report (PDF) · TechRadar · CyberScoop
Commentary
The convergence of AI tools and state-sponsored disinformation is an increasingly concerning trend, and the specificity of this campaign is notable. Targeting data centers — the physical infrastructure that makes AI possible — suggests a strategic interest in slowing U.S. AI compute buildout by weaponizing domestic opposition. The “concerned citizen” personas generated by ChatGPT itself add an uncomfortable irony to the whole situation.
That said, the “minimal impact” caveat matters. These campaigns consistently underperform because AI-generated astroturf still lacks the authentic community engagement that drives real political action. The bigger risk isn’t the current crop of clumsy influence ops — it’s what happens when these techniques improve. OpenAI deserves credit for publishing detailed threat reports on misuse of its own platform, but the fact that their tools are being used to generate anti-AI propaganda is a narrative loop that practically writes itself.