PRC-linked influence operations are targeting AI debates in the US
OpenAI
OpenAI caught China-linked accounts using AI to push fake narratives about US tech policy, data centers, and tariffs. They're also spreading false claims about ChatGPT itself—AI fighting AI in the propaganda war.
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 dropped a report this week that reads less like a corporate blog post and more like a counterintelligence briefing. The company says it identified and shut down a cluster of accounts linked to the People's Republic of China that were using its own tools to manufacture social media content aimed at American audiences. The targets weren't random. These operations zeroed in on the exact fault lines of US tech policy right now: data center buildouts, tariff fights, and the broader anxiety around AI's economic footprint.
What makes this different from the usual bot-farm playbook is the meta layer. Some of the content didn't just push political talking points—it spread fabricated claims about ChatGPT's capabilities and behavior, seemingly to sow distrust in the tool itself or muddy public understanding of what OpenAI's models can and can't do. That's a new wrinkle. Influence campaigns used to target elections and social wedge issues. Now they're apparently targeting the credibility of the AI companies whose products are becoming infrastructure for how millions of people get information.
OpenAI didn't publish exhaustive account counts or engagement metrics in what's been shared so far, but the framing is clear: state-linked actors are treating AI generation tools as force multipliers for narrative manipulation, and they're doing it fast enough to latch onto news cycles about tariffs and data center construction almost in real time. That speed is the tell. Manual troll farms take time to spin up talking points. AI-assisted ones can pivot to whatever's trending within hours.
The timing lands awkwardly for an industry already fighting perception battles on multiple fronts—energy use, job displacement, safety regulation. Adding foreign influence operations to that pile means AI companies are now doing double duty as both the subject of political debate and, unwillingly, a tool used to shape that debate. OpenAI positioning itself as the one calling this out is notable too; it's simultaneously an admission that its platform got used this way and a signal that it's willing to publicly attribute the source.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI safety nobody wanted to build for but everyone should've expected: not rogue AGI, but garden-variety state propaganda wearing an AI-generated mask. I don't think this makes the case for locking models behind more walls—if anything it argues for better provenance and detection tooling that open ecosystems can actually inspect and improve. Closed labs quietly catching and disclosing this stuff is fine, but it shouldn't be the only line of defense, because right now it mostly relies on OpenAI grading its own homework.
Read more about this at: OpenAI
