China-linked influence campaigns are shaping AI discourse in the United States

If you follow how AI tools are being used — and misused — in online influence campaigns, a new OpenAI report offers a rare inside look. The company has documented several covert operations linked to actors in China that used its models to generate content aimed at U.S. audiences. The campaigns touched on debates over American technology policy, narratives around data centers, tariff discussions, and even false claims about ChatGPT itself. Here's what the report found and why it matters.

What the Campaigns Were Trying to Do

According to OpenAI's findings, the operations behind this activity appear connected to actors based in the People's Republic of China. Rather than a single effort, the report describes a set of coordinated campaigns, each with a different focus.

One campaign targeted ongoing U.S. debates about technology, pushing content designed to influence how Americans think about key issues in the tech sector. Another centered on data centers — a topic that has become increasingly contentious as AI infrastructure expands and communities debate the costs and benefits of large-scale computing facilities. The goal seemed to be shaping public narratives around these facilities.

A third effort focused on tariffs, injecting content into discussions about U.S. trade policy. And in a notable twist, one campaign spread false claims about ChatGPT itself, spreading inaccurate stories about the very tool being used to produce the content.

How AI Was Used in These Operations

The report makes clear that OpenAI's models were used to help generate text for these campaigns — essentially serving as a writing assistant for content meant to look organic. However, the company notes that the campaigns did not achieve significant engagement or viral traction. In other words, while the actors were active and persistent, their content largely failed to reach a wide audience.

This pattern matches what independent researchers and OpenAI have found in previous threat reports: AI can speed up content creation and lower the barrier for running influence operations, but it doesn't automatically make that content persuasive or widely shared. Building a real audience still requires distribution networks, and those remain a bottleneck.

Why False Claims About ChatGPT Stand Out

Perhaps the most interesting element of the report is that one of the campaigns spread misinformation about ChatGPT itself. This isn't entirely unprecedented — influence actors have previously spread rumors about the platforms and tools they use — but it highlights a strange dynamic in the AI era: the same tools that enable influence campaigns can also become targets of disinformation.

For everyday users, this is a useful reminder to be skeptical of viral claims about AI tools, including dramatic stories about what these systems supposedly do or don't do. Checking whether a claim traces back to a credible source — like the company's own documentation or reputable news outlets — is a good habit.

What This Means for the Broader Picture

OpenAI regularly publishes these threat intelligence reports as part of its effort to be transparent about how its models are misused. The company says it disrupts accounts engaged in this kind of activity when it detects them.

The bigger takeaway is that AI-linked influence operations are becoming routine rather than exceptional. Multiple countries' actors have been documented using generative AI for covert messaging aimed at foreign audiences. The good news from this report is that the campaigns' actual impact appears to have been limited. The content was produced, but it didn't spread widely — suggesting that detection and disruption, combined with the natural difficulty of building authentic-seeming audiences, can keep these efforts from gaining real momentum.

For readers who want to understand the full details, OpenAI's report is worth reading directly, as it includes more specifics about the campaigns' tactics and targets.

For regular internet users, the practical lesson is simple: stay alert to coordinated messaging on hot-button topics like tech policy, data centers, and trade — especially when posts seem formulaic or push the same talking points across many accounts.