An AI-generated clothing model on Shein’s site bore an uncanny resemblance to alleged murderer Luigi Mangione, confirmed by facial recognition tools.
The shirt sold out before Shein pulled the listing, blaming a third-party vendor for the fiasco.
The incident adds to a growing list of AI-generated ad disasters, from deepfake influencers to celebrity likeness lawsuits.
The marketing team at Shein probably thought they’d found the perfect model: chiseled jawline, brooding eyes, the kind of face that could sell low-quality shirts to anyone scrolling at 2 a.m. There was just one problem—their AI-generated hunk was Luigi Mangione, the 26-year-old charged with killing UnitedHealthcare’s CEO last December.
The shirt, a basic cotton one, appeared on Shein’s website with what AI researcher Henk van Ess identified as a likely Midjourney creation. Facial recognition tools confirmed the match. Within hours of going viral, the listing vanished faster than Shein’s return policy fine print.
“We are aware of the product and have removed it from our site,” a Shein spokesperson told TMZ. The company blamed an third-party vendor, shirking blame without identifying the supplier in question. The shirt reportedly sold out before anyone at Shein noticed their new model’s resemblance to the most wanted man in America last week.
But of course, you can access the advertisement if you visit the Wayback Machine.
Shein promised “appropriate action” against the vendor, which probably means finding a supplier who can generate fake models that don’t accidentally replicate people on FBI watchlists.
The incident sparked the usual X pile-on, with users torn between horror and grudging admiration for whatever algorithm thought “alleged CEO killer” was the right aesthetic for budget casualwear.
the fact that your face can be used, without consent, to promote products that you’ve never even touched thanks to genAI…
add in the fact that they’re using the image of an incarcerated individual who has already lost all autonomy.
Shein’s algorithmic face-plant joins a growing gallery of AI mishaps, including Scarlett Johansson, who practically made herself the fake face of the anti-deepfake movement.
The famed actress threatened to sue OpenAI over a suspiciously familiar voice assistant. She then sued Lisa AI for making an AI companion using her image. And later, her likeness was included without her consent in a fully AI-generated campaign featuring different Jewish artists wearing a t-shirt giving Kanye West the middle finger.
An AI-generated version of Scarlett Johansson featured in the viral video decrying Kanye West’s antisemitism. Image: Instagram
Just reecently, gaming company Nexon got caught using AI influencers in TikTok ads for The First Descendant.
“These AI ads make me sad and worried for the future of the game,” one Redditor wrote. “The worst part of this all is that whoever is in control of the Official TFD TikTok accounts is aware of the backlash against the AI ads and has gone out of their way to ignore and hide it. They have been deleting and reuploading the AI videos whenever they get too much backlash in the comments.”
Regulators are scrambling to catch up, but so far, regulation is crawling while AI sprints.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
