Hacker News Reader

Comment by CPUHall | Hacker News Reader

Title is misleading. The ruling said training on books is fair use. What got them in trouble was grabbing the books from LibGen instead of just buying them, which is a pretty dumb way to lose $1.5B. If I torrented half a million books I'd be getting a lot more than a $3k-per-book settlement.

View on HN

Anthropic also claimed that those pirated books had not been used to train their models.

They've since bought vast numbers of second hand books and scanned those for training, which appears be legal.

We went through this with Google. Scanning books is legal as long as you then immediately destroy them. Then since training with material you own is fair use they’re good to go.

> which appears be legal.

Interesting phrasing from you, Simon. Is there any legal theory under which it should be illegal?

Appears to be legal in that current rulings and legislation say that to do so does not run afoul of fair use.

LLMs can reproduce the training data verbatim, under some circumstances. They have to be trained not to.

No idea, I'm not a lawyer.

Thanks, I’ve changed the submission title from the author’s original.

"The ruling said training on books is fair use."

Like when people bring this up as if it's somehow just like acceptable.

It's not at all. That's utter bullshit. The people that run the courts have been bought out. They have no idea what the far-reaching consequences of this is.

Reply on news.ycombinator.com