Wading through the detail of the EU AI Act, I was taken by the definition it offers in Article 5(1) for the kind of AI practices that will be prohibited.
On a quick first read I was struck by the idea that surely social media algorithms, well documented to be designed with the purpose of maximising engagement as a core objective, would fall within the definition.
On a further read, and having sought out some discussion of the issues, I came across this article that sets out a very interesting timeline of the draft of the relevant provisions.
How the text of the current Act introduces uncertainty is a fascinating expression of how these things work and it’ll be worth following how the implementation of these provisions operate, or not, to balance the interests of developers and users of AI.
Thus, one might expect that manipulative social media AI practices would be broadly prohibited within Article 5 of the EU's proposed regulation.
www.bloomberglaw.com/...

