Problem child EU AI Act

“I am very concerned about the EU AI Act that was finalized tonight. I see the EU as a self-proclaimed pioneer, but in protectionism. The copyright issue has eclipsed the basic model criticality classification issue! Scandalous!

We will see what kind of migration or hurdles we will face due to the bureaucratic jungle of documentation, technical proofs and database “transparency” – some will be geopolitically pleased. Either it will be a fighting exit of the young industry or some will just pull out the likes of Aleph Alpha, which could be their salvation. In the USA, there were only gentle decrees. And the proposal of the three countries was surprisingly late, and also (as I learned internally) largely the initiative of France. I am saddened that Germany did not come out on top, although I thought Habeck’s separation between technology and application was clever.

I see another opportunity: the issue of AI liability is still largely unresolved globally (see also the dire consequences and the backpedalling of the test pilot projects in LA/San Francisco). By clarifying liability, limiting liability and sandboxes, we would even have a chance to attract the AI industry worldwide as a location factor before the EU directive on liability, which is also being prepared, comes into force!”