The product of that analysis does not contain the data itself, and so is not a violation of copyright.
That’s your opinion, not the opinion of a court or legislature. LLM products are directly derived from and dependent upon the training data, so it is positively considered a derivative work. However, whether it’s considered sufficiently transformative, or whether it passes the fair use test, has not to my knowledge been determined in court. (Note that I am assuming US law here.)
The courts have yet to come to a conclusion, the lawsuits are still ongoing. I think it’s unlikely they’ll conclude that the models contain the data, however, because it’s objectively not true.
The clearest demonstration I can think of to illustrate this is the old Stable Diffusion 1.5 model. It was trained on the LAION 5B dataset, which (as the “5B” indicates) contained 5 billion images. The resulting model was 1.83 gigabytes. So if it’s compressing images and storing them inside the model it’d somehow need to fit ~2.7 images per byte. This is, simply, impossible.
You’ve got your definition of “derivative work” wrong. It does indeed need to contain copyrightable elements of another work for it to be a derivative work.
If I took a copy of Harry Potter, reduced it to a fine slurry, and then made a paper mache sculpture out of it, that’s not a derivative work. None of the copyrightable elements of the book survived.
Because that would be sufficiently transformative, and passes all the fair use tests with flying colors.
If you cut up the book into paragraphs, sentences, and phrases, and rearranged them to make and sell your own books, then you are likely to fail each of the four tests.
But even if you manage to cut those pieces up so fine that you can’t necessarily tell where they come from in the source material, there is enough contained in the output that it is clearly drawing directly on source material.
If you cut up the book into paragraphs, sentences, and phrases, and rearranged them to make and sell your own books, then you are likely to fail each of the four tests.
Ah, the “collage machine” description of how generative AI supposedly works.
It doesn’t.
But even if you manage to cut those pieces up so fine that you can’t necessarily tell where they come from in the source material, there is enough contained in the output that it is clearly drawing directly on source material.
If you can’t tell where they “came from” then you can’t prove that they’re copied. If you can’t prove they’re copied you can’t win a copyright lawsuit in a court of law.
That’s your opinion, not the opinion of a court or legislature. LLM products are directly derived from and dependent upon the training data, so it is positively considered a derivative work. However, whether it’s considered sufficiently transformative, or whether it passes the fair use test, has not to my knowledge been determined in court. (Note that I am assuming US law here.)
The courts have yet to come to a conclusion, the lawsuits are still ongoing. I think it’s unlikely they’ll conclude that the models contain the data, however, because it’s objectively not true.
The clearest demonstration I can think of to illustrate this is the old Stable Diffusion 1.5 model. It was trained on the LAION 5B dataset, which (as the “5B” indicates) contained 5 billion images. The resulting model was 1.83 gigabytes. So if it’s compressing images and storing them inside the model it’d somehow need to fit ~2.7 images per byte. This is, simply, impossible.
That’s not in question. It doesn’t need to contain the training data to be a derivative work, and therefore a potential infringement.
You’ve got your definition of “derivative work” wrong. It does indeed need to contain copyrightable elements of another work for it to be a derivative work.
If I took a copy of Harry Potter, reduced it to a fine slurry, and then made a paper mache sculpture out of it, that’s not a derivative work. None of the copyrightable elements of the book survived.
Because that would be sufficiently transformative, and passes all the fair use tests with flying colors.
If you cut up the book into paragraphs, sentences, and phrases, and rearranged them to make and sell your own books, then you are likely to fail each of the four tests.
But even if you manage to cut those pieces up so fine that you can’t necessarily tell where they come from in the source material, there is enough contained in the output that it is clearly drawing directly on source material.
Ah, the “collage machine” description of how generative AI supposedly works.
It doesn’t.
If you can’t tell where they “came from” then you can’t prove that they’re copied. If you can’t prove they’re copied you can’t win a copyright lawsuit in a court of law.