A wave of fake soldiers is marching across social media, and none of them are real. Using image generators and video tools, creators are building convincing military personas that gather large audiences before the accounts are unmasked and removed. Researchers who track the trend say the goal is almost always the same: to turn the credibility of a uniform into money or influence.
An audience built on a lie
One account, presenting an American service member who did not exist, drew more than a million followers over a few months before it was taken down, funneling its audience toward paid adult content, Military.com reported. Others have paired AI-generated faces with political messaging aimed at specific audiences. Watchdogs have taken to calling the phenomenon "digital stolen valor," the online equivalent of wearing medals one never earned, borrowing the authority of military service to win sympathy, trust or a sale.
Scams and propaganda
The tactic serves several ends. Fraud investigators say synthetic soldiers are being used to run romance scams, striking up online relationships before asking for money; US regulators have logged tens of thousands of romance-scam complaints and billions of dollars in reported losses in recent years, though how much of that now involves AI is hard to pin down.
The same techniques are being used for political propaganda. In one case documented by the Ukrainian outlet Euromaidan Press, AI-generated videos posing as Ukrainian soldiers pushed demoralizing messages to Ukrainian audiences, one clip gathering hundreds of thousands of views. Separately, researchers cited by Prism News found AI-generated war footage spread across several platforms and accounts, together drawing tens of millions of views before removal. The claims about who is behind such campaigns vary and are often hard to verify independently; what is clear is that the tools to make them are now cheap and widely available.
How to spot them, and why it's getting harder
The fakes still leave clues. Guides such as one from PBS News point to telltale signs, hands with the wrong number of fingers, objects that warp or vanish, and a too-perfect sheen to the lighting, along with details a real service member would get right, like the markings on a uniform. Reverse image searches can sometimes reveal an account's origins. But the technology is improving quickly, and safeguards are imperfect: digital watermarks meant to flag AI content can often be stripped away, so their absence proves nothing.
The platforms' problem
Social networks require creators to label AI-generated content and remove accounts that break the rules, but enforcement leans heavily on self-disclosure, which a scammer has no reason to provide. In practice, moderation is mostly reactive: accounts are usually flagged and pulled only after they have already reached a big audience. As generating a believable fake person becomes a matter of a few prompts, the burden is shifting onto viewers to doubt what they see, and onto platforms to detect deception at a scale, and speed, they have so far struggled to match.
