Your Meeting Room AI Has Been Grading Your Face. Europe Just Made It a Crime.

That smart camera in your last video meeting? In Europe, if it was reading your face to guess whether you looked bored, tired, or unconvinced — that's now illegal. Not frowned upon. Not against company policy. Against the law. And that ban kicked in back in February of last year.

If you've ever sat in a meeting and wondered

If you've ever sat in a meeting and wondered whether the new camera was doing more than just framing the shot — you weren't being paranoid. Europe's new A.I. rules just drew a hard line between a meeting tool that helps you and one that quietly grades you. The European Union's A.I. Act now bans what's called emotion recognition in the workplace. That means any system that tries to read your feelings from your face, your voice, or your body — that's prohibited. So where's the line between a helpful assistant and a covert judge? That question runs through this whole story.

Start with the everyday version of this. You join a call. There's a smart camera tracking who's in the room. Maybe it adjusts the lights, maybe it follows whoever's talking. Under the new rules, that's mostly fine — it's logistics. But the moment that same system starts scoring how engaged you look, or flagging that you seem tired, or rating how confident you sounded — it crosses into banned territory. One layer keeps the meeting running. The other layer evaluates the humans in it. Europe just made that second layer illegal.

Now, how common is this really? According to a survey by the O.E.C.D. — that's the international group that studies economic policy — more than seven in ten managers said their company already uses at least one automated tool to instruct, monitor, or evaluate workers. So this isn't some far-off scenario. Most organizations already have this technology sitting in scope. If you've been on a work call this year, there's a real chance one of these systems was in the room with you.

There's also a rule most workers never heard about

There's also a rule most workers never heard about. Before a company deploys this kind of high-risk A.I., it has to tell worker representatives and the affected employees. Silent rollout — quietly switching it on and saying nothing — that's no longer allowed. For you, that means the days of finding out by accident are supposed to be over.

And the penalties have teeth. Companies that break these rules can face fines up to thirty-five million euros — or up to seven percent of their global revenue, whichever is bigger. That's not a parking ticket. That's a number that ends up in a boardroom conversation. The deadline for the high-risk employment systems to fully comply lands in August of next year.

Here's the catch the experts keep pointing to. The hard part isn't understanding the law. It's that many vendors built their products with the helpful part and the judging part tangled together. Pulling out the behavioral analysis costs real money. So companies face a choice — downgrade to a plain transcript tool, or pay to rip the evaluation engine out.