The Truth About New Influence Campaigns

Influence campaigns are changing how we interact online. A newly uncovered operation shows exactly how covert actors try to hijack our daily debates, using AI tools to blend in with everyday Americans while targeting domestic discussions through a series of digital coordination efforts that likely originated in China. But the operators didn't try to invent new arguments from scratch. Instead, they jumped onto existing, highly sensitive domestic topics to fan the flames of public anger.

The activity centered on two distinct clusters of digital manipulation. One campaign, dubbed Data Center Bandwagon, focused on convincing Americans that the rapid expansion of physical data centers is driving up their local electricity bills. It's a clever trick. The second campaign focused on the highly controversial topic of trade tariffs, framing them as a sneaky way for countries to control global technology. But both campaigns shared a common goal: to blend into the background of authentic domestic political debates and push them toward division. So don't be fooled.

Inside the digital playbook

How did these operators attempt to pull this off? They relied heavily on ChatGPT to generate their material. By feeding specific prompts to the AI, they created social media comments, written posts, and custom imagery designed to look like organic American commentary. But there was a catch in their instructions. The prompts explicitly told the AI to focus all criticism on U.S. President Donald Trump regarding tariffs, while completely ignoring Chinese President Xi Jinping, who has also used tariffs heavily.

Let's break down the technical details of how they operated. They weren't just casually browsing the web. The technical footprint of the operation reveals a highly structured setup, and the actors used specific technical workarounds and localized methods to hide their true identity. Here's the deal.

• The anti-data center campaign was linked to an unnamed Chinese technology company that holds active contracts with regional Chinese governments.
• Operators used Virtual Private Networks, commonly known as VPNs, to bypass geographic restrictions and hide their actual locations.
• The actors prompted the AI tool in simplified Chinese but demanded outputs in both English and Chinese.
• They created fake profiles on major social media platforms, specifically targeting users on X and YouTube while pretending to be ordinary Americans.
• The tariff campaign expanded its reach by generating short comments and comics in multiple languages, including English, Italian, Japanese, and traditional Chinese.

But the operators didn't stop at public posts. They used AI to improve their own internal operations instead. So here's the strange twist: the actors fed their own internal work reports into the AI to edit them, and those reports contained sensitive operational security details about their social media campaigns, detailing their goals of building highly credible, long term accounts that could evade platform detection.

Faking the metrics of success

They uploaded documents to the AI asking how to best exploit Facebook's systems. It was strategic. They wanted to know how to use groups, pages, hashtags, advertising tools, and recommendation algorithms to spread their message, and they also sought ways to bypass security systems designed to detect coordinated inauthentic behavior. But it couldn't be that simple.

They targeted security researchers on X with a fake campaign. But the network blasted out claims alleging that a massive user data breach had occurred, a claim that Nimmo said "never happened." That's a lie. This shows the group was willing to invent crises out of thin air when piggybacking on real news failed.

Why these campaigns failed to land

The high-tech tools were impressive. But these influence campaigns didn't get the reaction the creators wanted,they failed completely. On the Brookings breakout scale, which measures the reach and impact of online operations, these campaigns only scored a 1 and a 2, showing they couldn't connect with real people. So the average internet user just ignored them.

Why did they flop? Quality control. It turns out that AI-generated propaganda still has a major problem, and many of the images used by the Chinese actors appeared clunky and awkward with text that relied on overly direct, aggressive messaging showing a clear lack of familiarity with English and internet culture nuances. They simply didn't know how to make a post go viral naturally.

"I do want to be really clear here: this was not a case of an influence operation creating a debate. The debate existed already. This was an influence operation from China trying to interfere in it. We didn't see any signs that it succeeded."

Ben Nimmo, OpenAI Principal Investigator

Ben Nimmo analyzed the activity. He said such views are "reasonable" and "sincerely held" by many participants. Many Americans hold sincere, deeply felt views on both sides of these issues, and we can't ignore that. But the real danger isn't the debate itself. It's the fact that a covert foreign entity tried to pose as patriotic citizens to manipulate the conversation, using the very AI technology they were publicly attacking.

What this means for your feed

So what does this mean for your inbox and your social media feeds? Stay skeptical. You must remain deeply suspicious of highly emotional posts on hot button issues, because these actors happily latch onto genuine grassroots movements and ride the wave of organic public anger. But they're not starting a new argument. It's far easier to piggyback on topics that already make you angry, and they want their operations to look highly effective to their backers , so don't let them fool you.

The campaigns likely originated in China. But investigators have stop short of directly blaming the Chinese government or official state actors for the operation. Many of the tactics, themes, and technical overlaps match known propaganda campaigns run by the Chinese state in the past, and as these tools become more accessible, the line between organic debate and foreign manipulation will only continue to blur.

How to spot the manipulation

Protecting yourself from these digital operations does not require advanced technical skills. It requires a healthy dose of skepticism. When you see a highly divisive post about energy bills, tech tariffs, or infrastructure, take a second look at the account posting it before you share or comment.

Look for the telltale signs of automation and foreign coordination. Accounts that post constantly but have almost no real followers, profiles that use weirdly formal or clunky English, and accounts that only post about a single controversial topic are major red flags. They're easy to spot. So by recognizing these patterns, you can keep your online conversations authentic and keep foreign operators out of your community debates,but it's not hard to stay sharp.