Is it distribution though if someone trains a model in switzerland through downloading copyrighted material, training AI on it and then distributing it...
Or what if not even distributing it but rather distributing the outputs of the LLM (so closed source LLM like anthropic)
I am genuinely curious as to if there is some gray area that might be exploited by AI companies as I am pretty sure that they don't want to pay 1.5B dollars yet still want to exploit the works of authors. (let's call a spade a spade)
Using copyrighted material to train AI is a legal grey zone. The nyt vs openAI case is litigating this. The anthropic settlement here is about how the material is obtained. If openAI wins their case and switzerland rules the same way I dont think there would be a problem
This might go down (I think) to be one of the most influential court cases to happen then.
We really are getting at some metaphysical / philosophical questions and maybe we will one day arrive at a question that just can't be answered (I think this is pretty close, right?) and then AI companies would do things freely without being accountable since sure you could take to the courts but how would you come to the decision...?
Another question though
So lets say that the nyt vs openAI case is going on, so in the meantime while they are litigating (lets say), could OpenAI still continue doing the same thing while the case is going on?