Federal judge rules copyrighted books are fair use for AI training
nbcnews.com
nbcnews.com
Now, with judicial opinions on this fair use firming up, I am hopeful this will allow them to train on every book they’ve ever acquired and release those models to the world.
Yay, more AI slop to pollute the internet with.
Edit: I respect your position to not engage. We can just do things regardless of those who would disagree (which is good imho when the risk of harm is low and the benefit potentially high), in this case while still adhering to the law as set forth by a court. The judicial interpretation guides that you can train if you own a copy of the work; build accordingly.
When Search Engine Services meet Large Language Models: Visions and Challenges - https://arxiv.org/html/2407.00128v1
I'm not going to engage with anyone who tries to force a weak comparison with an LLM to a different kind of technology. I think we as developers and scientists are well aware that these LLMs are sufficiently different from what's come before for such a comparison to work.
The problem with the Internet Archive was it jumped into uncontrolled lending. Basically, there was no practical reason to buy a book that the Internet Archive would "lend" you. That simply isn't true for an LLM citing a book, which may still cause me to read it.
[section removed; refer to harshreality's context in sibling comment, which is more thorough than what I had commented]
As a reasonable person who believes one should be reasonable until they can no longer be reasonable, as a hacker, this is a green light to be unreasonable as it relates to onerous copyright regulations infringing on the commons. Copyright stakeholders took too much from the commons, and this enables people to take back (with a combination of judicial review and technology).
https://www.library.upenn.edu/news/hachette-v-internet-archi...
https://blog.archive.org/tag/controlled-digital-lending/
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
This is the moment their liability accelerated. It was stupid, impulsive and put--and may continue to put--the whole project at risk.
Edit (wrt your reply): Drop into a weekly lunch at the SF location and have a conversation on this. Additional context is always helpful, I find.
[1] https://blog.librarylaw.com/librarylaw/2007/07/internet-arch...
What they allow doesn't matter. Under controlled lending, the Archive was operating within precedent. I'm not against launching a test case for uncontrolled lending. But doing it with the entire library, irrespective of publication date or jurisdiction, and through the main organisation was just stupid.
I love the Archive. And as I said above, I'm not fundamentally against uncontrolled lending as a legal test case. But doing it with the entire catalogue, instead of just e.g. older books, and doing it out of the main organisation such that there is no legal segregation between the experiment and the entire project struck me as incredibly impulsive. They're wonderful and weird. But perhaps not the best steward of what's pitched as an archive.
1. Almost all physical libraries, including at universities, had shut down, and IA's efforts were to compensate for that. Supply of previously-purchased-for-public-use books had dried up overnight.
2. Normal 1:1 lending was not observed, but DRM was still applied. No books lent during that period were usable outside of a narrow window, without the end user intentionally circumventing the DRM which is a distinct crime and isn't trivial for the average computer user.
3. IA entered into dialogue with major academic publishers before implementing the emergency library. They published an opt-out address on their blog. That wasn't as visible to all publishers or independent authors as it probably should have been, but everyone knows if it had been opt-in instead, authors and publishers all would have reflexively refused, despite the unique situation. Mainstream publishers and most authors simply do not care if people can't access library books for a month or two. They would ban libraries if they could. They're so high on their intellectual property rights they think those can't be relaxed no matter what.
4. As far as I know, all NEL books were scanned PDFs, from donated library book collections. Casual readers don't read scanned books for entertainment. The idea that this significantly displaced kindle purchases, for instance, or lending of real ebooks from local libraries, needs a citation.
There's plenty of additional background here: https://blog.archive.org/national-emergency-library/
As long as they're not violating copyright laws in output, it's fine and good.
> Despite siding with the AI company on fair use, Alsup wrote that Anthropic will still face trial for the pirated copies it used to create its massive central library of books used to train AI.
You've never read an opinion that required creativity?
Yes. Judges in a common-law system don't just "apply the law," they literally make law. Treating judges as automatons fundamentally misunderstands their role, which makes any predictions based on that misunderstanding likely specious.
And I don't just base my predictions on that understanding, it's more of a pass-through thing. I base my predictions on observed reality, where there is a legislative branch that makes laws and a judicial branch that interprets them within narrow confines (supposedly, they basically have free reign to go wild and say up is down and down is up, but that rarely happens because they don't want to be laughed at in public)
Think Blu-Ray DRM but for more than video, it's already happened with publishers and college textbooks.
In my view, one real gray area is in image/video generation, especially "x in the style of y" kinds of shenanigans. As a society we may need to consider some better protections for an artist's/studio's style, otherwise distinct and novel and interesting styles will become watered down into a sea of bland mimicry until the sweet release of the heat death of the universe.
it's not a gray area when humans do it, so it's not a gray area
well over 50% of those obnoxiously loud anti-ai artists make a living off fan art, and until 3 years ago, copyright concerns would be scoffed at
More smug nonsense emanating from Misanthropic. There is no creativity that is enabled. People tweak the prompts like children until something that was stolen from others emerges.
Most people working on "AI" have never created anything substantial. They rely on utilizing other people's creations. It is very sad that Alsup caves to big tech and issues a vibe ruling.
Edit: I see another commenter, presumably human, clarified: "legally-acquired copyrighted books" Even with the arguments about AI being potentially helpful to disabled humans, one healthier route is to help each other out directly instead of dividing and conquering with technology, in the name of helping. Feels like one of the aims of Capitalists is to put us each into our Matrix (1999 movie) battery capsules and bleed us dry while we're distracted.