[wouterjanl]

Article 5 (f) of the EU AI Act, fresh from the press (published two weeks ago), prohibits "the (...) use of AI systems to infer emotions of a natural person in the areas of workplace and education institutions, except where the use of the AI system is intended to be put in place or into the market for medical or safety reasons". [1]

Penalties consist of fines up to EUR 35M or 7% of an undertaking's turnover, whichever is higher (article 99, §3 AI Act).

The prohibition will apply only from 2 February 2025 (see article 113 (a) AI Act).

Not a day too early, arguably even too late. Governments of the world, follow suit!

[1] https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L...

[sofixa]

But I was told the EU was just overregulating AI and no innovation and etc...

I have to say I'm impressed by how quickly the EU managed to hash out AI regulations. I've only skimmed it but there are multiple good things there, on top of the above, like bans on critical decisions being taken only by AI, the need to have humans in the loop, the ability to be able to explain AI decisions for important things, etc.

It will definitely have kinks to iron out, but it's a good start and shows what good AI regulations could look like.

[rsynnott]

The EC will see these regulations as essentially a first draft; expect a new, likely stricter set in a few years, and probably on a regular cadence after that. We're currently on the Fifth AML Directive, for instance (the first was in 1990).