When AI doppelgängers come for you: 5 real-life horror stories
In folklore, meeting your doppelgänger was always a bad omen. An exact copy of you, wandering around somewhere, was thought to be a warning of impending death. But at least those sinister doubles belonged to horror stories.
Today, digital doppelgängers are very real, and you don’t even have to encounter your own to be in trouble. A familiar face in an ad could be an AI impostor trying to scam you. Meanwhile, a copy of you could be promoting products you've never heard of, speaking in your voice, or chatting with your friends long after you're dead. All it might take to create one is a handful of photos, videos, or voice recordings, often collected without your knowledge or consent.
From stolen faces to digital ghosts, here are five real-life AI horror stories that make the old legends seem almost comforting.
1. A face here, a body there: Fashion’s AI Frankenstein experiment
We’ve all been warned that AI is coming for our jobs. But for models and other people whose livelihoods depend on their appearance, losing work might be just the tip of the iceberg.
According to a recent Reuters investigation, draft contracts from brands including Diesel and Calvin Klein seek permission to put models’ faces on other people's bodies, swap faces between models, and alter their age, skin and features. The contracts could give brands broad control over how models’ images are used, including for AI training. In other words, models could help train the very technology threatening their jobs.
And the danger extends well beyond modelling. Deepfakes may still seem like a celebrity problem, but INTERPOL warns that what was “once seen as a threat mainly to public figures” now increasingly targets ordinary people. And unlike models, who can at least challenge the fine print, ordinary people may never have agreed to anything in the first place.
2. Your favorite influencer’s evil twin wants your money
In August 2026, a UK social media creator claimed on Reddit that someone had copied their face, voice, tattoos, personal belongings, and even speech patterns. The AI double pretended to work in the same field, and urged followers to sign up for an app. The fake account reportedly amassed 100,000 Instagram followers and blocked the real creator when they tried warning people.
Earlier in 2026, lifestyle influencer Emily Schuman found an Instagram ad falsely showing her promoting weight-loss drugs. After angry followers flooded her inbox with questions, she discovered more posts where her AI double pushed cosmetics and blood-testing services. When contacted, the blood-testing company said a scammer was behind the post featuring its brand.
Scammers have been long hijacking trusted social-media accounts to push fake ads that lead to phishing sites. Now they don’t even need to hijack an account. Making a convincing-looking AI double is enough. An ad blocker can help keep scam ads off your screen, while AdGuard’s Browsing Security can warn you about known phishing sites. But neither can tell you if that familiar face in an ad is actually real. So the rule of thumb is to check the creator’s other channels and visit the advertiser's website directly rather than clicking the ad.
3. Resurrecting the dead requires only a subscription now
Forget séances and Ouija boards. These days, bringing back a dead relative might take just a few minutes of video and a (paid) app.
In November 2025, startup 2wai went viral with an ad showing a pregnant woman chatting on her phone with an AI replica of her deceased mother. Her son grows up talking to the digital grandma and eventually tells her he’s about to become a father himself. She never ages, and the entire relationship rests on a three-minute recording made before her death. The ad sparked a backlash, with critics accusing 2wai of exploiting grief. Its CEO confirmed plans to charge for conversation time, promising that unpaid bills wouldn’t mean deleting grandma. But terms can change. The premise is practically begging for its own Black Mirror episode: a boy grows up with the grandmother he never met, only to lose her when the app shuts down.
Not everyone finds digital resurrection entirely horrifying, though. But there’s always the question of who controls a deceased person’s likeness, memories and voice, and what happens to the intimate conversations relatives have with their replica. We’ve explored these questions in our article on the death-tech industry and digital life after death. And for those who’d rather decide what happens to their online identity while they’re still alive, our TechTok guide to planning your digital afterlife covers the practical side.
4. After you die, Meta’s AI could keep answering your DMs
In horror films, ghosts haunt old houses. In the AI era, they might haunt your Instagram comments and DMs instead.
Meta has actually patented technology that could keep your social-media presence alive long after you're gone. Filed in 2023 and granted in December 2025, the patent describes someone’s absence from social media as “affecting the user experience,” with the impact becoming “much more severe and permanent if that user is deceased.” Meta’s proposed solution is an AI trained on your posts, comments, likes and other activity, capable of interacting with friends and even simulating voice or video calls. The patent suggests that “other users may not notice an absence.”
There are situations where such a feature, while ethically questionable, might come in handy. An influencer taking a break could let an AI double handle routine interactions with followers. But Meta explicitly brought up death, which makes its reasoning rather bizarre. Getting an unexpected DM from your dead friend could arguably be far more distressing than their absence. Worse, you might keep chatting with their AI double for months before finding out they’d died. In all fairness, the patent says AI-generated responses would be identified as such, but that hardly makes the prospect less unsettling. Meta insists it has no plans to develop this technology. Still, there’s no guarantee the idea will stay buried.
5. Make a deal with the devil your doppelgänger — before someone else does
Not every digital double is created against its owner’s wishes. Some people are voluntarily putting their AI replicas to work.
In September 2026, Australian footballer Cristian Volpato signed a deal with California-based Callandor, giving the company exclusive rights to his digital likeness, voice, biometric data and performance stats. His AI twin could appear in ads, speak different languages, and even give fans personalized football tips, all while the real Volpato gets on with his life.
Compared with having your identity stolen, this sounds like the lesser evil. Volpato agreed to the deal and could make money from it. Others are taking precautions: Matthew McConaughey has even trademarked clips of himself and his famous “Alright, alright, alright” to fight unauthorized AI copies. But handing a company exclusive rights to your face, voice and biometric data is still a big step, especially when nobody knows where this technology will go next. Perhaps the modern equivalent of selling your soul to the devil is simply signing a licensing agreement for your digital twin.
The privacy dilemma behind the mask
Behind these stories lies a more familiar threat: the collection and exploitation of personal data. Digital doubles can be built from faces, voices, biometric information, online activity, and even private conversations and memories. That information may be stored, shared, or used to train AI models, sometimes without the knowledge or consent of the people involved. And while a fake profile or video might be taken down, the underlying data and copies can be much harder to erase. Unlike a Halloween costume, your face and voice aren’t things you can simply take off when the party is over.
