wikipedia states: Generally, copying cannot be proven without some evidence of access; however, in the seminal case on striking similarity, Arnstein v. Porter, the Second Circuit stated that even absent a finding of access, copying can be established when the similarities between two works are "so striking as to preclude the possibility that the plaintiff and defendant independently arrived at the same result."
This is a different situation in which exact replication can be reasonably occurred without access to the original.
Secondly, can you actually claim Github has violated copyright if it doesn't have any claims to the work in question?
I think it's totally plausible that they win this in the long run.
1) So you are saying if I get a disk duplication machine I can freely copy and distribute blu ray disks as long as I don't watch the movie on the disk?
2,3) Seems pretty settled at this point, look at the cases around the VCR and copy machine. In general the one using the machine is liable. The creator of the machine can be held liable if there aren't substantial non infringing uses.
> It's not a violation of copyright to train a model.
Many people on HN assert this based on the Authors Guild vs. Google case, but it's quite important to keep in mind that that case was about Google creating a search algorithm, which is not generating "new" output.
We are talking about a very different kind of system here and in many other cases. Claiming the Authors Guild case sets precedent for these very different systems seems unbased to me.
> It's not a violation of copyright to train a model.
This is a very bold assumption, one that I assume will not hold in the court of law in all cases. I think the nuanced question is: to train a model that does what, exactly.
Let's say distributing meth recipes is illegal[1], can one legally side-step that by training a model that spits out the meth recipe instead? No court will bother with the distinction, causation is well-trod ground.
1. As an example - not sure if its illegal. You may replace with classified nuclear weapon schematics if you like.
It's not illegal to train a model to spit out classified nuclear weapon schematics. Possessing the original data might be. Releasing software that does this might be illegal, but not for copyright reasons, which is the issue at hand.
