@kkarhan@infosec.space@YoSoyFreeman and if you read closely I don't think any apply.
It's just that learning and scraping can't be banned nor criminalized without literally breaking the entire world - because guess what your #Fediverse client and the servers on both our instances instances do:
They'll inevitably cache and store those contents regardless of whether or not I'd provide them with the rights to do so explicitly, because they are "technically necessary" and thus don't have to ask for permission.
Still, by the logic of yours, people would be as much subject to copyrights as programs and thus we'd still criminalize any learning from unlicensed sources...
Needless to say I'm not gonna defend #GAFAMs and other rentseeking asshole corporations, because I think #AI should be build upon #transparency and #openness and that we should feed it with permissively licensed source material only...
In fact, the lack of the latter and the absurdly high computational cost of building even a single neuronal net is why I don't even consider even trying to make one.