EU AI Law - A Sham

Before anybody gets too excited over the EU AI act, I recommend reading article 5 of the artificial intelligence act. There are so many exceptions for governments as to make the whole act completely useless.

AI legislation must protect citizens from their governments first. Instead the EU sees themselves as the righteous party that is above reproach while private governments and citizens need to be reigned in.

The reality is that if governments see a chance to build AI systems that work in their favor and to the detriment of free citizens, they will build them, and they will use them.

You can already witness this in the emerging legislations for digital communication and age restrictions on the Internet which not only allow but implicitly mandate the practices that are outlawed by the AI act, specifically to use AI to identify and classify users and their communication.

Another example is the newly proposed "Medienstaatsvertrag" in Germany, which will allow the so-called "Medienanstalten" to use any technical means to spy on everything people publish on the internet. The internet platforms will be forced to provide interfaces for those "technical means" to access all necessary data. While the proposed law does not specifically mention artificial intelligence one does not need much imagination to figure out that AI is exactly what they will be using.

AI legislation should instead focus on three areas:

  • protection of citizens against government use of AI (prohibit governments to use AI for anything that has to do with privacy, free speech and surveillance of citizens)

  • transparency (citizens must know at all times when they or their data is subject to AI)

  • Create the conditions for a vibrant market where citizens have a real choice between services that use more or less AI (this is probably the same as ensuring complete transparency)

comments powered by Disqus