‘AI Homeless Man’ Photo Trend on TikTok is Wasting Police Time in the US and UK
A trend for AI-generating strangers inside people’s homes has prompted cross-border police forces to issue warnings.
The Oak Harbor Police Department (OHPD) in Washington addressed a false social media post that claimed a homeless person visited Oak Harbor High School.
“This claim is not accurate,” the OHPD said this week, per Fox 13 Seattle. “There have been no incidents or safety concerns involving the homeless population at any Oak Harbor School District campus.”
Meanwhile, Dorset Police in England received a call from an “extremely concerned parent” who believed that a man was inside their family home with their daughter.
It was actually a prank, but police said it used “valuable deployable resources” for the hoax. “If you receive a message and pictures similar to the above antics from friends or family, please attempt to check it isn’t a prank before dialling 999,” Dorset Police said.
What is Happening?
Both of these incidents have been caused by AI-generated pranks that have mainly been happening on TikTok and are created via Google’s new Nano Banana AI model that was recently integrated into Adobe Photoshop.
The prank started as the “AI homeless man” trend, which sees people add a photo of an unhoused person to a photo of their home. Google’s Nano Banana — officially known as Gemini Flash 2.5 — is a powerful photo editor that can be controlled using natural language prompts. It means that basically anyone who knows how to use an AI image model can now edit photos the way only professionals could just a few years ago.
The cruel joke has since extended to girlfriends pranking their boyfriends that a shirtless maintenance man is in their house, or that men in balaclavas are inside the house.
@holleecheyenne The funniest part about this is that we have cameras in the house that he could’ve checked at any point in time. 😂 #marriage #humor #prank #bluecollar #funny ♬ Boccherini Minuet Trio – Aura Classica
While the pranks are clearly in bad taste, they are often on parents or people who are not that digitally literate. Much of the population remains unaware of just how easy it has become to fabricate images like this, an alarming state of affairs that the police have somehow been dragged into.
And in case that wasn’t bad enough, OpenAI’s video generator Sora 2 has been running rampant with an avalanche of deepfake videos across the internet depicting everything from Jake Paul announcing he is gay to Michael Jackson working in Walmart.
Image credits: Header photo licensed via Depositphotos.
