Together with Dr. David Bomhard and Prof. Dr. Patrick Glauner, I wrote an op-ed for the German Tagesspiegel Background Digitalisierung & KI that was published today.
We claim that the AI Act does not fix existing problems that hamper innovation in AI but instead creates many new obstacles. Moreover, we are illustrating that EU policy-making nowadays features serious structural deficits, leading to laws that are overstepping EU competences and overlooking Better Regulation principles.
How does this article fit to my rather positive previous messages on the AI Act here on LinkedIN? Firstly, this is a 'Brandbrief' that addresses general problems within EU policy-making and goes beyond the AI Act. Secondly, I remain convinced that the AI Act is conceptually wrong and still belief that the approach proposed by the Legal Committee was superior (a. EU AI bills of rights, b. adjustments in sectorial legislation, c. AI strict liability framework). The AI Act tries to create a horizontal 'silver bullet' applicable to all AI systems - a legislative approach that failed numerous times before. The law is also obsessed about risks (82 of 85 articles) and overlooks almost completely how to promote innovation in AI (3 articles + coordinated plan that is barely implemented). Thirdly, the European Parliament is on a good way to improve the law significantly and seems to fix at least some of the problems that Patrick, David and I have identified in our article.
To sum up, the AI Act is IMHO the wrong approach to regulate a broad technology category such as AI. However, I also acknowledge that it is not possible to withdrawal the whole proposal and replace it with a new law at this stage. Therefore, my colleagues and I, at the technical/political level, are doing everything in the European Parliament to improve the text and make it work.
Read the full article here.