Lawsuit Against EleutherAI for Books3 [pdf]

> Because the LLMs cannot function without the expressive information extracted from the Infringed Works and are retained inside the Books3 database and the LLMs, these datasets and models are themselves infringing derivative works, made without Plaintiffs’ permission and in violation of their exclusive rights under the Copyright Act.

The implications of this are staggering if they win. It means no one will ever be allowed to release any training data, ever. Not in the way that OpenAI, Midjourney, or any of the other big players have been training their models. Takeaway: it will become impossible to make open source LLMs, because all training data is by definition copyright infringement and not fair use.

> Plaintiffs and the members of the Class have been injured by Defendants’ acts of direct copyright infringement. Plaintiffs and the Class are entitled to statutory damages, actual damages, restitution of profits, and other remedies provided by law.

They keep asserting they’ve been damaged. I’d love to see them sit down and calculate how much damage has been done to any individual author. I sure hope everyone arguing for copyright enforcement is happy being protected from “damages” that can likely be measured in cents.