Facial Comparison: Compare Two Faces After 150 Fake Photos

A UK secondary school got an email. Attached were images that used real students' faces, pulled from ordinary class photos, but the bodies and scenes were fake. Made by AI. And there were 150 of them, all counted as child sexual abuse material under UK law. The gang wanted money to keep them from spreading. This already happened. It's not a hypothetical. And it's the reason a boring-sounding tool called facial comparison (a way to check how much two faces actually match, using measurements of things like the distance between your eyes or the shape of your jaw) just became something every parent should understand.

TL;DR: Facial comparison tech, the kind that checks whether two photos show the same real person, is becoming the fastest way for schools and families to tell whether a threatening AI image is fabricated or actually shows their child in danger.

Here's the math that should scare you more than the headline did. A survey behind the "In Plain Sight" report found that 49 percent of UK headteachers say their school has already dealt with some kind of incident involving a student's photo being misused, altered, or used to extort money. Only 43 percent say they'd actually know what to do if it happened to them. That gap, between "this is happening" and "we're ready for it," is the whole story.

Face comparison and facial features are the missing step in every school deepfake blackmail case

Think about what actually happens when a school gets one of these threats. Someone, usually a parent or a teacher, opens an email with attachments and a deadline. The pictures look real enough to make your stomach drop. But are they? Nobody in that room, not the head teacher, not the school's IT person, not even the police officer who eventually gets called, has a fast, reliable way to answer that question on the spot. That's the actual emergency. Not just that fake images exist, but that nobody can quickly tell fake from real when it counts most.

This is where facial comparison earns its keep. At its simplest, it's software that takes two photos, maps out facial landmarks (specific points like the corners of your eyes, the tip of your nose, the edge of your jaw), and produces a match score, basically a similarity score showing how likely it is that both photos show the same actual person. It won't undo the horror of a fabricated image existing. But it gives investigators, and eventually families, a way to establish facts fast instead of guessing under pressure while a criminal's deadline ticks down. Good face verification also means fewer wasted hours chasing an image that a proper facial comparison check could have cleared, or confirmed, in minutes.

Criminals are scraping ordinary photos from school websites and social media, feeding them through AI tools to manufacture abuse images, then demanding payment to keep the results offline. This article is part of a series, start with Social Media Identity Verification Macron Eyes Id Scanning.

reporting summarized from Yahoo News Canada, citing the "In Plain Sight" report by Aidos

And here's the part that should really bother you: right now, this is still a manual process for the criminals. Someone has to actually go find the photos, download them, run them through the deepfake tools. According to Malwarebytes, experts warn this could soon get automated, letting criminals scrape names and faces off school websites at scale, no human effort required. If that happens, this stops being a rare, targeted attack and starts being something closer to spam. Except the spam is child abuse imagery with your kid's face on it.

How facial comparison actually works when a school photo is in question

You upload a reference photo (a confirmed, real image, like last year's official school photo) and the photo in question. The software analyzes facial features, things like the spacing of the eyes, the curve of the mouth, the shape of the chin, and generates a percentage showing how similar the two faces are. It's not magic. It's math applied to bone structure. But it's a real answer, in minutes, instead of a guess made in a panic. This is the same face comparing process, sometimes called a face compare, that lets a school run a facepair check against a verified photo rather than guessing from a screenshot.

What schools got wrong before AI made facial recognition and face recognition threats personal

Most school safeguarding policies were written for a world where a "bad photo" meant something embarrassing got posted, not something manufactured from scratch. According to TechRound, while 85 percent of schools do some kind of risk assessment before publishing student photos online, only 43 percent of those assessments actually consider the risk of AI misuse or deepfake generation. The rules exist. They're just built for a threat that's a decade out of date.

Meanwhile the criminal side of this is scaling fast. Child sexual extortion reports rose 19 percent in the first half of 2024 compared to the year before, according to data referenced by TES, with a staggering 2,600 percent jump specifically in reports involving girls. That's not a slow drift. That's a criminal business model finding its footing, and it's exactly why facial recognition and face verification checks are moving from "nice to have" to standard practice for any school handling a photo-based threat.

Can facial comparison and face comparison tell if a photo is really you or an AI fake?

Mostly, yes, but with a catch worth knowing. Facial comparison technology is built to answer "is this the same person," not "was this image staged or generated by AI." So a tool can tell you, with a percentage, how much two faces look alike across two photos. What it can't always catch on its own is whether the scene around that face was built by a generator. That's why the smartest use of this tech pairs a facial comparison check with AI-detection review, not either one alone.

People also ask: does a match score prove an image is authentic?

No, and this matters. A high match score means the face in the photo likely belongs to a specific real person. It does not mean the surrounding image, the pose, the setting, the body, is real. AI can graft a real face onto a completely fabricated scene. Facial comparison confirms identity. It doesn't confirm the scene actually happened. Families and investigators need to understand that distinction before they draw conclusions from a percentage alone. Previously in this series: Biometric Device Quality Dhs Bets 440m On Cameras Podcast.

School deepfake blackmail is forcing a rethink of what "proof" even means for face comparison

Look, nobody's saying facial comparison solves this crisis by itself. It doesn't stop a gang from scraping a photo off a school's "meet the staff" page in the first place. But it does something that matters just as much in the moment of crisis: it gives a scared parent and a panicking school an actual answer instead of a shrug. Here's where it gets interesting though. The same tool that identifies "yes, that's your child's face" can also help confirm the opposite: "no, that's not actually your daughter, that's a different kid's photo run through a filter." Both answers matter. Both change what happens next.

If you've ever wondered whether a photo or profile claiming to be someone is really who it says it is, that's the exact question this kind of technology exists to answer. One thing you can actually do, before any of this ever touches your family: know that a reverse image search or a facial comparison check against a verified reference photo (a real, confirmed picture you control, like a recent yearbook shot) is a legitimate, fast first step if you ever receive a suspicious image claiming to show your child. Don't forward it. Don't post it anywhere trying to "prove" it's fake. Save it, document when you got it, and get a verified comparison done quietly, ideally with law enforcement in the loop. Some services now offer this as an API, so a school's IT team can run a face comparison check against a trusted photo without manually uploading files one at a time, which matters when a threat arrives with dozens of images attached.

How does an online face comparison tool actually generate a result?

You choose two photos, a trusted reference and the one you're questioning. The system maps facial features and landmarks, calculates distances and proportions, then compares two faces mathematically. Within seconds it instantly shows a percentage, an instant AI similarity score representing how alike the two faces are. It's not a guess. It's pattern math applied to bone structure, the same basic idea humans use when they say "that doesn't look like my kid," just made precise enough to write down and hand to a police officer.

Ninety-three percent of headteachers already know AI can turn ordinary school website photos into abuse material, according to the "In Plain Sight" survey findings reported by TechRound. Knowing isn't the same as being ready. The gap between those two things is currently being filled by criminal gangs, not by policy. Every day that gap stays open is a day some family gets a message they were never trained to handle, and a quick facial comparison and face verification step is one of the few concrete things a school can put in place before that day arrives.

So here's the actual question worth sitting with tonight: if your kid's school got one of these emails tomorrow, with fifty images attached and a countdown clock, would anyone in that building know to reach for a fast, documented answer, or would they just reach for the "delete" button and hope the problem goes away quietly? Because quiet is exactly what these gangs are counting on. Up next: Social Media Identity Verification Macron Eyes Id Scanning P.

facial comparison: Frequently Asked Questions