I'd like to see a discussion around these aspects. I see only cries re big platforms and news outlets.I'm neither of these.

Smaller platforms are exempted on the directive (ie, startups, single person corporations, etc.) as well as anything that doesn't do profit oriented content/active content moderation.

But arXiv is not small, it has 1.4 million e-prints, is far bigger than most of other scientific platforms. It is non-commercial though.

Why not? AFAIK about a 1/3 of the arXiv articles are later published in journals.

Why would you? The Link Tax and Upload Filter specifically affect service providers that moderate user content to generate profit, if you upload something with explicit consent of all rightholders then this is explicitly exempted from the directive.

So? They're not published there illegally, arXiv (AFAICT) is non-commercial and publishing similar content elsewhere does not retroactively make the existing content illegal.

Sure, they are not published illegally. IANAL but as I understand the process, is like this: the author submits to arXiv and either (a) gives arXiv a perpetual NON-EXCLUSIVE licence to distribute (which does not change the fact that the copyright is with the author) or (b) the author chooses a CC-BY licence. The author submits the article to Journal and after acceptance, may transfer the copyright to Journal. Is perfectly legal and it works like this for a significant part of the arXiv articles. See also here in the comments for cases when a revised version of the article from Journal is posted on arXiv. But now, with the new EU Copyright Directive, how will this delicate process interact with the dumb one-size-fits-all automatic filters which may detect (wrongly) that the article on arXiv infringes the copyright of the Journal. What if arXiv will receive a bombardment of requests from various Journals? What will they do? They are admirable but they don't have the surface to fight this ddossing from journals. Or maybe my concerns are void, I'd be very happy if this is the case.

> What if arXiv will receive a bombardment of requests from various Journals?

Why are journals not sending a bombardment of requests already? Why would they start now?