[bryanlyon]

These lawsuits seem to be missing the point. Copyright protects their right to copies and the piracy sites are OBVIOUSLY at fault here, but there is a good argument that OpenAI's work in transformative and that they're not liable for copyright violations.

One argument this law firm has made is that ChatGPT can summarize the books, so it must have read them. This is spurious and meaningless as many sources summarize books and some like cliff notes actively sell summaries and analysis of books!

In the end, I think we need to follow Japan and say that AI training is transformative and not a copyright issue.

One clarification: That is training an AI where the training data isn't returned directly, but instead it works off of it.