CaraComp
CaraComp
Forensic-Grade AI Face Recognition for:
Get Started7-day refund guarantee**
Podcast

Biometric Camera: A Facial Match Is a Lead, Not Proof

Biometric Camera: A Facial Match Is a Lead, Not Proof

Quick answer

Is a facial recognition match enough to arrest someone?

No. A facial recognition result is a probability score showing two faces look alike, not an identity. It can point investigators toward a person, but other evidence must support it. New York City police policy says a match alone cannot justify a stop, an arrest or a search warrant.

Biometric Camera: A Facial Match Is a Lead, Not Proof

0:00-0:00

This episode is based on our article:

Read the full article →

Biometric Camera: A Facial Match Is a Lead, Not Proof

Full Episode Transcript


Police in the United States have wrongfully arrested at least fourteen people because a facial recognition match turned out to be wrong. Now Scotland is considering this same technology, and it may do so without a single national law to govern it.


If you've ever had your face caught on a doorbell

If you've ever had your face caught on a doorbell camera, a store camera, or a street camera, this story is about you. Most of us never agreed to be searchable. And it's unsettling to think a computer's guess could put you in handcuffs for a crime in a town you've never visited.

Here's what happened. Professor William Webster oversees biometrics and surveillance cameras in England and Wales. He told the outlet FutureScot that Scotland is in a "perilous position." Police Scotland is weighing live facial recognition. That means cameras that scan faces in a crowd as people walk by. But Scotland has no primary legislation for it. That means no full law passed by lawmakers. So the real question is this. When a computer says "this might be the person," who decides what happens next?

Start with what Webster says is missing. Scotland has a Code of Practice. That's a set of guidelines. But Webster says it may not even apply to police use of Age Verification Face Scan 2 Year Miss Lets 16 Pass As 18. And he calls it "a poor substitute" for a real law. A guideline can be ignored. A law has consequences.


Trusted by Investigators Worldwide
Run Forensic-Grade Comparisons in Seconds
Detailed facial comparison reports. Results in seconds.
Get Started
7-day refund guarantee**

Why does that matter

Why does that matter? Because of what a match actually is. A facial recognition result isn't an identity. It's a probability score. It says two faces look similar. Officials fear some officers treat that score as hard evidence. That's how an innocent person ends up arrested.

Look at the New York City police department's own written policy. It says a face match alone can't justify a stop. It can't justify an arrest. It can't even justify a search warrant. Officers need other evidence first. That's the right rule. The trouble is whether people follow it.

The A.C.L.U. has documented that they often don't. In case after case, police had written warnings that a result was only a lead. People were still arrested. Some spent months in custody. The Detroit case of Robert Williams is one of the best known. The warning was printed right there, and it didn't stop the arrest.


Then there's the bias problem

Then there's the bias problem. Researchers found that when they changed the race of the faces being compared, false matches jumped. For darker skinned people, those false positive rates were often a hundred times higher. UK regulators have also warned there aren't enough safeguards to make sure these systems aren't biased against particular groups. So the risk isn't spread evenly. Some people face far more of it than others.

There's a fair counterpoint, and it matters. The technology has improved. In recent UK deployments, when the match threshold was set high enough, there were no false positives at all. Some experts argue the answer isn't to ban the tool. It's to require real human checks. That means independent evidence and a documented trail of every search.

But that trail is exactly what's missing. Police aren't even logging their use of retrospective facial recognition. That's when they run a photo through a database after the fact. So someone could become a suspect, and no record would show why. For investigators, that's a process failure. For the rest of us, it means you might never learn why you were ever flagged.


The Bottom Line

The software isn't the main problem. At high settings, it can be quite accurate. The problem is people treating a strong lead like a finished answer. And without a law, nothing forces anyone to stop and check.

Face matching can point police toward someone. It can't prove that person did anything. The danger starts when a guess gets treated like a fact. Whether you investigate cases or just walk past a camera on your way to work, a match should be the start of a question, not the end of one. The full story's in the description if you want the deep dive.

Ready for forensic-grade facial comparison?

Full forensic reports with detailed similarity scoring. Results in seconds.

Run My First Search