That Kiosk Isn't Just Checking Your Face — It's Guessing Your Mood

There's a machine at some airports and stores that doesn't just check who you are. It tries to guess how you're feeling. And according to the Dutch Data Protection Authority, the science behind that guess is shaky enough that regulators just banned it outright in workplaces and schools.

You've probably unlocked your phone with your face

Now, you've probably unlocked your phone with your face and thought nothing of it. That kind of technology answers a simple factual question — is this the right person, yes or no. But a growing number of systems are doing something very different. They're looking at your face and claiming to know your mood. If that feels invasive, you're not wrong to feel that way. The good news is that once you understand the difference between those two things, you'll see exactly why lawmakers are treating them so differently. So what actually separates confirming your identity from reading your emotions?

Let me start with the distinction, because it's the whole ballgame. When a system confirms your identity, it's doing a lookup. It takes your face and compares it against a stored photo — a match, or no match. That's evidence. Either it lines up or it doesn't. Emotion inference is a completely different animal. The system takes your facial expression, maybe your voice tone, maybe your posture — and then it deduces a feeling. It's not matching. It's guessing.

Picture a guard at the airport. If he checks your face against your passport photo, that's solid — the faces match or they don't. Now picture that same guard staring at your expression, deciding you look suspicious, and pulling you aside for questioning. One is based on evidence. The other is based on an assumption about what your face means. And assumptions can be wrong.

The Dutch Data Protection Authority put it bluntly

The Dutch Data Protection Authority put it bluntly. Emotion recognition rests on contested assumptions, carries high error rates, and brings real risk of discrimination. The problem is that these systems present a subjective guess as if it were hard fact. A screen that flashes "angry" makes it look certain. But the technology can't actually deliver that certainty. Your expression means different things across cultures, moods, and moments. For you, that means a machine could label you frustrated when you're just tired — and treat you accordingly.

Now, a lot of people assume that if a company just tells you it's using emotion technology, the problem's handled. That feels reasonable. We're used to thinking that disclosure equals safety. But here's why that logic breaks. A warning label on a broken tool doesn't make the tool work. Telling you a faulty system is running doesn't make its guesses accurate. That's exactly why European lawmakers didn't stop at "just disclose it." Under the A.I. Act, emotion recognition is banned in workplaces and schools completely. The question there isn't how you disclose it. It's whether you're allowed to use it at all.

And there's a quieter loophole worth knowing. Under current privacy rules, an employer running an emotion system on a job candidate isn't required to tell that person which specific emotions it detected. So a worker might know a system is watching — but never learn it flagged them as nervous or dishonest. That gap builds false confidence on both sides.