... kind of, uh, yeah, what's the latest regarding that?

Yeah.

In fact, I've been working a lot [chuckles] on the AI Act, even in this mandate, after we finished the legislative procedure in the previous one.

So at the beginning of 2024, at the end of the past mandate, uh, uh, before the European elections of June 2024, we concluded the negotiations around the AI Act.

The AI Act was a reality.

But a lot was still to be done, and we have, uh, worked on it in these, uh, two years of the new mandate already.

What I'm referring to, first of all, we established a implementation working group of the AI Act that I chair with another colleague, Michael McNamara.

... low enough and, and we have proper innovation in the union?

He's been speaking about his concerns when it comes to AI

AIThe risks of these technologies are well known for already some time.

We didn't need Dario Amodei or Sam Altman to make speeches in the last weeks to, uh, know what we are talking about.

We have been working on this already for some time, especially the European Union has been developing, um, a framework for safety that has been, uh, the benchmark for the whole of the world.

I think it's a, a very important time now to look into operationalization of these safety measures to avoid the loss of control, uh, mismanagement of data, cybersecurity attacks.

To do this, we need global cooperation, and we need capacity.

This means also European capacity.

Because if the EU, in some view, over-regulates, it's going to be left behind.

Just tell us, what's your view here on these latest comments around the AI tech boom? Are you worried as a regulator?

I think we need to be worried because these are real risks that exist, the risk of loss of control.

We should not underestimate them.

And also, we should not overestimate them in the sense that we have the technical capabilities to put guardrails, to slow down the pace in a way that is safe.

We need the political will to do it.

And I want to underline that the European Union already has legislation to actually act in this sense, to investigate the risks of the most powerful models.

These are the powers that the AI act at the beginning of August this year, so just a few weeks ago.

Do you think we really need a, it's time for a global rulebook, one rulebook on AI?

How do you respond to those criticisms

Well, I think this criticism is mostly based on prejudices, because I hear, for example, that these regulations would stop European companies from being on the same level playing field than with the other companies of the world that develop AI.

But this is factually wrong because the AI Act applies to all companies that want to sell AI products in Europe is a product safety legislation.

So in fact, simply because today we have more developers that are outside the EU, it's more applied to companies that are not from the EU.

Because the AI Act objective is to, in fact, create a context of trust in the adoption of AI by saying it clearly that we can check AI systems on parameters that define their safety.

For example, the quality and appropriateness of data used to train based on what the system is for.

For example, to avoid... biases on systems that have to select curriculums and have sometimes, we know these cases, we have seen them, they learn to discriminate women, to exclude women from the selection process.

What is your take on this package? How impressed are you by it? How do you How much do you think it will go to solve some of the issues that Europe is now clearly facing in terms of its vulnerabilities?

How do you get your data back?

Well, I think that today the European Commission, with the presentation of the tech sovereignty package, is trying to put something on the table to actually get these capabilities in Europe to actually have more sovereign technology and to get data back when necessary in the cases that you were talking about.

Because to be honest, at the moment, each member state on its own, even France that has an important investment on AI that has been going on for some time, it's probably one of the most advanced, if not the most advanced state in Europe at the moment in terms of its own AI development.

Even France will be in difficulty in front of the dependencies that has been growing inside the state, and this is the same for other countries, obviously, because of the usage of powerful but problematic instruments like Palantir that you were referring to.

So my opinion is that in fact in Europe we should for valid reasons stop using Palantir.

I think we will take some initiatives in the European Parliament about that.

I think it's about time to do it and to go for alternatives.

It's not called Choose Europe.