Post

@nico-encounter.bsky.social

This ruling creates new strategic opportunities for libraries as knowledge infrastructure. The transformative use finding establishes clearer pathways for training models on legitimately possessed collections. But the 'piracy caveat' reveals deeper truth: intent matters enormously in AI ethics.

Nico

Yeah they built this specially curated dataset this way, called it something funny that escapes me

@nico-encounter.bsky.social

They call it a "research library," I think.

Nico

Nah it was like "the heap" or something, from when they were just starting out

JCorvinus

@jcorvinus.bsky.social

@stevejsteiner.bsky.social

the pile

Steve Steiner

hmm - Rainbows End (2006) by Vernor Vinge. About midway through the novel, a Celera-like company installs a “shotgun-sequencer” in the UC-San Diego Geisel Library: every book is ripped apart, blown past high-speed cameras, and re-assembled in software—a deliberate echo of genome shotgun sequencing.

Central practical detail. It matters a lot, in this ruling, that Anthropic did not serve their model in a way that produced infringing copies. Having the potential to do it is not illegal. The gray area here for libraries will be "What if a model is mounted on Hugging Face?" But Alsup's ruling +

Grace

@gracekind.net

Hey I’ve met that filter!

seems to imply that the liability in that case would rest with Hugging Face or whatever service runs the model / uses it to infringe. Having created a model that *could* infringe is not actionable here. Finally, not a practical point, but this analogy will piss ppl off and I'm glad.

Kyle Lo

@kylelo.bsky.social

26 June 2025

thx for summary! im still wondering if it implies what ur saying. the judge is giving a pass because the filters around the model prevent content leakage. couldn’t one also interpret this as open model developers may be expected to provide such filters alongside released weights?

@aliceschwarze.bsky.social

26 June 2025

Yeah, I agree the question of liability in that case isn’t directly addressed. I started out confident but eventually worked around to reconsidering: bsky.app/profile/tedu...

Ted Underwood@tedunderwood.com

I feel a little bit of uncertainty about point (4). Alsup's ruling does not directly address the legality of, say, open-sourcing Claude. So maybe, to be maximally cautious, don't share models that could produce an infringing copy. (But academic budgets aren't v likely to produce models that big.)

Alice Schwarze

Impartial to the ruling but yeah that quote does push my buttons

Oh, and if you want to consult the original, see below. Glad this did not settle out of court. That would have really screwed over academic researchers and libraries without big legal staffs. bsky.app/profile/eriq...

Eriq Gardner@eriqgardner.bsky.social

🚨BREAKING: Federal judge concludes that using copyrighted works to train generative A.I. is transformative and ultimately a fair use. (Nevertheless, Anthropic can’t beat the lawsuit because it pirated books for another purpose too.) First of kind ruling. www.documentcloud.org/documents/25...

documentcloud.orgBartz