TikTok already knows what videos you like, what you search for, and how long you spend watching certain content. Now, a software developer claims the popular social media app is also judging how attractive you are.
According to developer Ayo Omolewa, who goes by Ayoships on X, TikTok uses artificial intelligence to analyze users’ faces, estimate their ages, identify their genders, and even determine their moods. And apparently, according to him, it also gives users an attractiveness score.
Omolewa says he discovered these things after reverse-engineering TikTok and extracting some facial-analysis models from the app. He then created a website called What TikTok Sees, which lets people upload photos and see how those AI models allegedly evaluate their faces.
Developer Claims TikTok’s Facial Recognition Technology Has Serious Biases
“I reverse-engineered TikTok,” Omolewa announced in a November 3 post on X.
He went on to say that he extracted several artificial intelligence models from the app and modified them to run in a web browser.
He claims that after doing so, he found TikTok “scores how attractive you are, guesses your age/gender, and calculates your mood with some of these on-device models.”
Somehow, even more troubling, Omolewa also claims the facial recognition technology has a significant bias against people with darker skin and older faces.
His website, supposedly using TikTok’s model, analyzes a photograph across four categories: attractiveness, estimated age, estimated gender, and mood. The site says it keeps uploaded photos on the user’s device.
All that being said, Ayo Omolewa’s findings haven’t been independently verified, and just because TikTok has facial-analysis models doesn’t mean the company uses them to decide what videos people see.
However, other researchers have previously raised concerns about TikTok’s facial-analysis technology.
In 2024, researchers at the University of Wisconsin-Madison found that TikTok could estimate people’s ages and genders using photographs, sometimes incorrectly. TikTok’s use of facial analysis, including for content recommendations, was also subject to a Canadian privacy probe.
Developer Also Makes Troubling Claims About Popular Mobile Games
His TikTok research isn’t Ayo Omolewa’s first look at what popular apps are allegedly doing behind the scenes.
He previously claimed Block Blast deliberately adjusts its difficulty so players have to watch more ads, according to a report by Mobilegamer.biz. Omolewa also said Candy Crush dynamically changes the difficulty depending on how the player plays, and Subway Surfers collects a ton of data points to personalize advertising and in-game offers.
Again, there has been no independent verification to support the allegations.
TikTok has, however, faced questions about how it handles personal information, including a lawsuit that claims it mishandled users’ private data. Whether Omolewa’s latest allegations have uncovered something even more sinister remains to be seen.