Facial Recognition: Your Face Is Already in the Database

It's a Friday afternoon in September of 2024, at Harvard Square T station, Boston subway. A man called Kashif Hoda is waiting for his train when a young man approaches him.

The young man is called AnhPhu Nguyen, and Hoda nor Nguyen knows each other. Hoda noticed the man's glasses, with their thick frame it was hard not to. What he did not know was that they were Ray-Ban Meta smart glasses, and the tiny little white light that indicates that they are recording, was something Hoda was unaware of at the time.

A few minutes pass. Hoda's train pulls into the station. The same man approaches him again.

"Do you happen to be the person working on minority stuff for Muslims in India?" the stranger asks. Hoda is stunned, he works in biotechnology now. But years ago, he was a journalist who wrote about marginalized communities in India. Articles buried somewhere in the internet's back catalog. Work he hasn't thought about in years.

"I've read your work before," the stranger continues.

They shake hands. The train doors are about to close. Hoda boards, confused but flattered. He doesn't get a chance to ask the obvious question: How do you know me?

A month later, a friend texts Hoda. He's in a viral video. Twenty million views and counting.

At the same station, same day, another stranger gets the treatment. A woman is waiting for her train. Caine Ardayfio, Nguyen's project partner, walks up, extends his hand.

"I think I met you through the Cambridge Community Foundation, right?"

She smiles. Stands up. Shakes his hand.

"Oh, yeah," she says.

They chat. Ardayfio mentions her work with the Foundation like he's been following it. References a project she was involved in. The kind of specific detail that only someone who actually knows you would remember. She's engaged, nodding along, probably trying to place his face in her memory of volunteer events.

Truth is, she never even meet Ardayfio before. The Cambridge Community Foundation connection? Scraped from the internet in real-time. Her volunteer work, her affiliations, her name, that specific project, all pulled from public databases while Ardayfio stood there making small talk.

She has no idea she's talking to a complete stranger who actually does not know her or her work. And she has no idea she's being recorded.

The smart glasses Nguyen wore that day start at $299 for the base model, which means all of this was accomplished using the cheapest option available.

Anyone can easily order them online. They look like normal glasses, thick frames, slightly nerdy, the kind you'd see on any college campus. The only thing that tells you these aren't ordinary glasses is a tiny white LED on the upper right corner of the frame that lights up when recording. But most people don't notice it. Hoda didn't.

Here's how they did it. Nguyen and Ardayfio built a phone app in just four days capable of all this. The glasses livestream video to Instagram, not publicly, so your friends, followers and stalkers won't get a notification, but as a way to feed footage to a computer program monitoring the stream. When the program detects a face, it captures the image and uploads it to PimEyes, a facial recognition search engine that crawls the public internet for matching faces. PimEyes returns websites where that face appears: old news articles, professional directories, archived blog posts.

An AI language model scrapes those sites for context: names, occupations, biographical details. Once it has a name, it queries people-search databases like FastPeopleSearch for addresses, phone numbers, names of relatives. All of this gets pushed to the phone app in roughly ninety seconds.

The information doesn't appear inside the glasses. It gets pushed to the phone app that Nguyen and Ardayfio built. So while Nguyen stood there looking at Hoda, making small talk, asking for directions, he was glancing down at his phone. Reading Hoda's entire biography in real time. His home address, old journalism work, family members. All of it scrolling across a screen while the person wearing the glasses nods and smiles and pretends they recognize you from somewhere.

The facial recognition industry is massive, and it's already deployed at scale. Clearview AI has scraped more than 50 billion images from across the web.

Facebook, Instagram, YouTube, Twitter, building a database that's been used by more than 2,400 US law enforcement agencies. US police have run nearly a million searches using Clearview since 2017. PimEyes, the tool Nguyen used, was founded in Poland in 2017 and is now run out of Georgia, but registered in the Seychelles, a tax haven where, despite a Data Protection Act passed in 2003, there is "virtually no enforced legal protection for personal data" as of 2025. A new Data Protection Act was passed in December 2023 claiming to align with GDPR standards but enforcement hasn't begun.

Between September 2024 and 2025, London's Metropolitan Police used live facial recognition more than 200 times, resulting in more than 1,400 arrests. The UK installed its first permanent facial recognition cameras on buildings and lampposts in Croydon during the summer of 2025. With 2,500 cameras, the borough of Hammersmith and Fulham has more cameras per person than any other place in the UK. Currently, 500 of these cameras are being upgraded with AI-powered facial recognition technology.

More than 200 million surveillance cameras with facial recognition technology are in use in China. Intersection cameras in Shenzhen employ facial recognition technology to detect jaywalkers, publicly humiliate them by projecting their faces onto screens, and lower their social credit scores by two to three points.

Facial recognition is of course not only used by law enforcement to fight crime.

In Russia, the facial recognition app FindFace launched in 2016 and quickly became a weapon for harassment. Users from the imageboard Dvach used FindFace to identify women who had appeared in pornographic films or advertised on escort service websites. After identifying the women, users shared copies of the women's VK social media pages and spammed the women's friends and family to inform them what they had discovered. In one case, a man used FindFace after discovering hidden-camera footage of a woman using a bathroom at a Moscow coffee shop. He identified her through the app and contacted her on VK to tell her he had the video.

Survivors of domestic violence are at risk too. Individuals seeking to remain hidden from former partners who threatened or engaged in domestic violence are now one reverse image search away from being found. Privacy rights group Big Brother Watch filed a legal complaint stating that tools like PimEyes enable "surveillance and stalking on a scale previously unimaginable" and represent "a nightmare for any victim of stalking, domestic violence or bullying."

Clearview AI was sued by the ACLU for endangering survivors of domestic violence, among others. The company's database of 50 billion images scraped off the web creates what critics call a "perpetual police lineup" for everyone, regardless of whether they've committed a crime.

PimEyes operates with even fewer restrictions. When asked how PimEyes would prevent abuse by noncustodial parents, the CEO suggested requesting a birth certificate or signed form. The company, however, admitted that it had only requested such documents twice and banned just one account. When confronted about dangerous uses, PimEyes responded that "our privacy policy prevents people from using our tool for this case." A policy is just words, and words are not a safeguard. Meanwhile, PimEyes sells subscriptions to anyone with an internet connection for $30 a month.

Facial recognition isn't just in the hands of law enforcement and stalkers.

It's rapidly crawling into everyday life. Australia now requires users to verify their age on social media platforms like Instagram, TikTok, Snapchat, Facebook, Reddit, and YouTube using facial recognition or age estimation technology. TikTok allows users to upload a photo ID along with three selfies for age verification, while Instagram verifies people through a short video selfie.

Facial recognition systems are operational at over 80 US airports, including Atlanta, Los Angeles, Denver, Chicago O'Hare, New York JFK, LaGuardia, Newark, San Francisco, Portland, Salt Lake City, and Seattle. Retail stores employ facial recognition to deter theft, analyze customer behavior, and improve store layout and marketing. Amazon forces delivery drivers to consent to biometric data collection via cameras in vehicles that track when drivers are distracted, not wearing seatbelts, or even yawning.

68% of users employ facial verification to unlock devices like their phones. 53% of the total US population engages with facial recognition systems regularly. The infrastructure is already here. The databases are already built and your face is already in them.