Meta Hit With New Author Copyright Lawsuit Over AI Training (reuters.com)
- Reference: 0175181477
- News link: https://yro.slashdot.org/story/24/10/02/2251230/meta-hit-with-new-author-copyright-lawsuit-over-ai-training
- Source link: https://www.reuters.com/legal/litigation/meta-hit-with-new-author-copyright-lawsuit-over-ai-training-2024-10-02/
> Farnsworth said in the lawsuit on Tuesday that Meta fed Llama, which powers its AI chatbots, thousands of pirated books to teach it how to respond to human prompts. Other authors including Ta-Nehisi Coates, former Arkansas governor Mike Huckabee and comedian Sarah Silverman have brought similar class-action claims against Meta in the same court over its alleged use of their books in AI training. [...] Several groups of copyright owners including writers, visual artists and music publishers have [3]sued major tech companies over the [4]unauthorized use of their work to [5]train generative AI systems . The companies have argued that their AI training is protected by the copyright doctrine of fair use and that the lawsuits threaten the burgeoning AI industry.
[1] https://fingfx.thomsonreuters.com/gfx/legaldocs/zgpoawemmpd/META%20AI%20COPYRIGHT%20LAWSUIT%20farnsworth.pdf
[2] https://www.reuters.com/legal/litigation/meta-hit-with-new-author-copyright-lawsuit-over-ai-training-2024-10-02/
[3] https://yro.slashdot.org/story/24/08/20/1524250/authors-sue-anthropic-for-copyright-infringement-over-ai-training
[4] https://yro.slashdot.org/story/24/08/14/223234/artists-claim-big-win-in-copyright-suit-fighting-ai-image-generators
[5] https://apple.slashdot.org/story/24/07/16/1443251/apple-nvidia-anthropic-used-thousands-of-swiped-youtube-videos-to-train-ai
Honeypot? (Score:1)
I wonder if it would be useful to set up some honeypots that would make robots think they've hit on a motherload of text documents, but were in fact being fed a trove of machine-generated text that looked plausible if a human skimmed a small portion, but as a whole taught any LLM a bunch of nonsense.
Re: (Score:2)
Heh, I bet Meta doesn't use its own troll farms.
Re: (Score:2)
As far as I know, "poisoning the well" is by far the best widely-available means to fight back against data harvesters. We owe them nothing except our ill will.
Plain Text passwords? (Score:2)
Was it not META that was storing passwords without even MD5? Storing them in plain-text? How embarrassing in 2024. Not even SHA512.
Scarcity based thinking (Score:2)
And Meta is probably the most close to OpenSource of the big models out there, why just sue them? Does the lawsuit split all damages evenly among all authors of the world? Didn't think so [1]https://www.genolve.com/design... [genolve.com]
[1] https://www.genolve.com/design/socialmedia/memes/writers-artists-want-compensation-for-training-the-LLM
They even admit it (Score:3)
If they can blatantly copy what they like then the average joe should be allowed to copy what we like too.
Re: (Score:2)
And my platform of choice - Usenet.
Re: They even admit it (Score:2)
You pretty much can.
What you cannot do is redistribute it.
If you want to ban this usage you should also take away degrees from anyone who pirated their books for university
Re: (Score:2)
An AI is redistribution.
Re: (Score:2)
Redistribution of knowledge, not of the copyrighted content. Designing a bridge after reading an engineering textbook doesn't make that bridge a redistribution of copyrighted material.
Re: (Score:3)
AI is just a delivery system, and yes, of copyrighted content too. Nothing more. Even if it was locked up for personal use only, it's still just a delivery system.
Re: (Score:2)
AI is no different to a search engine.
Re: (Score:2)
I'm pretty sure it's OK to read whatever you want as long as you buy or legally borrow a copy of the book. In fact you can read that book all you want, and then if you bought it, you could even loan it to a friend. Websites that we open to all, also OK to read. I hope you spend your time wisely now that you know all this material is available to you. Good health to you.
Re: (Score:2)
Tell that to the public libraries, who are battling this very issue. They can't seem to get a break from copyright infringement.
Re: They even admit it (Score:2)
17 USC 106(4): "to perform the copyrighted work publicly"
So yes.