The EU AI Act forces you to label a lot of AI generated content like images or video or text - mostly as a way to fight mis/disinformation. For images for example you do not have to do that if they are just illustrative illustrations but you have to do it if anyone could mistake it for something real (like a generated picture of a person). I kinda wonder how that applies to code (software). The code is not illustrative but functional and "a person did it" carries a lot of weight in regards to assumptions about quality so publishing unmarked slop code feels like breaking the spirit of the AI act. Does anyone have any legal guidelines about that?