Tomorrow, the EU will present its long-awaited proposal for laws for artificial intelligence.

Brace yourself for headlines à la “Europe gets tough on the AI industry."

But is it?
(Thread)
Some backdrop: The laws could make Europe the first world region with a dedicated rulebook for AI.

Officials hope that by coming up with rules early, they will set a global standard for AI regulation.
I’ve been following the genesis of the rules closely.

In my former job as Politico’s AI correspondent, I reported about every step that led to the document released tomorrow.
Last week, my former colleagues leaked a draft (which is now circulating online).

If it's what the EU will present tomorrow, two comments:

a) Looks like Brussels, indeed, wants to get tough on the AI industry.
b) But I’m not buying it, yet. https://drive.google.com/file/d/1ZaBPsfor_aHKNeeyXxk9uJfTru747EOn/view
Here’s why:

While the proposed rules look tough at first glance, they get more wishy-washy the more you decipher the bureaucratic jargon.

The EU, for instance, pitches “conformity assessments” — stress tests, if you will — for AI technology that poses high risks to users.
That sounds good. Problem is that the document does not clearly define what exactly qualifies an AI system as “high risk.”

That’s just the kind of loopholes the tech industry is waiting for.
The EU also suggests to ban some AI technology.

Wow, you’d think.

But then you take a closer look, for instance at this definition:
“AI systems used for indiscriminate surveillance applied in a generalised manner to all natural persons without differentiation.”

And even I can tell you that it seems super-difficult to ban anything based on a definition *that* vague (and I’m not a lawyer).
My point is: Once you get to the bottom of the document, you realise that the rules proposed tomorrow are a political gesture, above all.
Expect Margrethe Vestager and the rest of the crew to stress how important it is to regulate AI, in order to safeguard democracy, etc pp.

(TBC, I have no reason to doubt that she’s serious about it. On the EU top level, she has IMHO the best understanding of tech.)
But at the end of the day, what's going to happen tomorrow really isn’t decisive.

It’s about what will happen next.
In the coming weeks and months, Big Tech — plus European tech companies — will send their armies of lobbyists to the Berlaymont and Avenue de Beaulieu.

They'll provide officials with “suggestions” how to tweak the fine print, to make sure rules won’t “stifle innovation” etc.
And yes: EU officials do listen to them. (On the working level even more so than on the political level, based on my experience.)

Which, per se, is fine.
But years of reporting on EU tech policy have taught me that this is where policy is made — not on stages like the one where Vestager etc. will present the AI laws tomorrow.
(By the way:

I’ve read some tweets saying that the EU “will pass” AI laws this week. It won’t.

The Commission is the EU’s executive branch. It makes proposals, which are then discussed by member states and the European Parliament. ...
... I spare you the details, but: Even if things go super-smoothly, nothing will become hard law until in another two years or so. Likely more.)
So, when you see the headlines tomorrow, don’t let them fool you:

Yes, Europe might get tough on the AI industry, and tomorrow could be a first step.

It could. If it will be one remains to be seen.

The fight over Europe’s AI laws is only just beginning.
You can follow @JanoschDelcker.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: