Yesterday, popular authors including John Grisham, Jonathan Franzen, George R.R. Martin, Jodi Picoult, and George Saunders joined the Authors Guild in suing OpenAI, alleging that training the company’s large language models (LLMs) used to power AI tools like ChatGPT on pirated versions of their books violates copyright laws and is “systematic theft on a mass scale.”

“Generative AI is a vast new field for Silicon Valley’s longstanding exploitation of content providers," Franzen said in a statement provided to Ars. "Authors should have the right to decide when their works are used to ‘train’ AI. If they choose to opt in, they should be appropriately compensated.”

OpenAI has previously argued against two lawsuits filed earlier this year by authors making similar claims that authors suing “misconceive the scope of copyright, failing to take into account the limitations and exceptions (including fair use) that properly leave room for innovations like the large language models now at the forefront of artificial intelligence.”

This latest complaint argued that OpenAI’s “LLMs endanger fiction writers’ ability to make a living, in that the LLMs allow anyone to generate—automatically and freely (or very cheaply)—texts that they would otherwise pay writers to create.”

Authors are also concerned that the LLMs fuel AI tools that “can spit out derivative works: material that is based on, mimics, summarizes, or paraphrases” their works, allegedly turning their works into “engines of” authors’ “own destruction” by harming the book market for them. Even worse, the complaint alleged, businesses are being built around opportunities to create allegedly derivative works:

Businesses are sprouting up to sell prompts that allow users to enter the world of an author’s books and create derivative stories within that world. For example, a business called Socialdraft offers long prompts that lead ChatGPT to engage in ‘conversations’ with popular fiction authors like Plaintiff Grisham, Plaintiff Martin, Margaret Atwood, Dan Brown, and others about their works, as well as prompts that promise to help customers ‘Craft Bestselling Books with AI.’

They claimed that OpenAI could have trained their LLMs exclusively on works in the public domain or paid authors “a reasonable licensing fee” but chose not to. Authors feel that without their copyrighted works, OpenAI “would have no commercial product with which to damage—if not usurp—the market for these professional authors’ works.”

“There is nothing fair about this,” the authors’ complaint said.

Their complaint noted that OpenAI chief executive Sam Altman claims that he shares their concerns, telling Congress that "creators deserve control over how their creations are used” and deserve to “benefit from this technology.” But, the claim adds, so far, Altman and OpenAI—which, claimants allege, “intend to earn billions of dollars” from their LLMs—have “proved unwilling to turn these words into actions.”

Saunders said that the lawsuit—which is a proposed class action estimated to include tens of thousands of authors, some of multiple works, where OpenAI could owe $150,000 per infringed work—was an “effort to nudge the tech world to make good on its frequent declarations that it is on the side of creativity.” He also said that stakes went beyond protecting authors’ works.

“Writers should be fairly compensated for their work,” Saunders said. "Fair compensation means that a person’s work is valued, plain and simple. This, in turn, tells the culture what to think of that work and the people who do it. And the work of the writer—the human imagination, struggling with reality, trying to discern virtue and responsibility within it—is essential to a functioning democracy.”

The authors’ complaint said that as more writers have reported being replaced by AI content-writing tools, more authors feel entitled to compensation from OpenAI. The Authors Guild told the court that 90 percent of authors responding to an internal survey from March 2023 “believe that writers should be compensated for the use of their work in ‘training’ AI.” On top of this, there are other threats, their complaint said, including that “ChatGPT is being used to generate low-quality ebooks, impersonating authors, and displacing human-authored books.”

Authors claimed that despite Altman’s public support for creators, OpenAI is intentionally harming creators, noting that OpenAI has admitted to training LLMs on copyrighted works and claiming that there’s evidence that OpenAI’s LLMs “ingested” their books “in their entireties.”

“Until very recently, ChatGPT could be prompted to return quotations of text from copyrighted books with a good degree of accuracy,” the complaint said. “Now, however, ChatGPT generally responds to such prompts with the statement, ‘I can’t provide verbatim excerpts from copyrighted texts.’”

To authors, this suggests that OpenAI is exercising more caution in the face of authors’ growing complaints, perhaps since authors have alleged that the LLMs were trained on pirated copies of their books. They’ve accused OpenAI of being “opaque” and refusing to discuss the sources of their LLMs’ data sets.

Authors have demanded a jury trial and asked a US district court in New York for a permanent injunction to prevent OpenAI’s alleged copyright infringement, claiming that if OpenAI’s LLMs continue to illegally leverage their works, they will lose licensing opportunities and risk being usurped in the book market.

Ars could not immediately reach OpenAI for comment. [Update: OpenAI’s spokesperson told Ars that “creative professionals around the world use ChatGPT as a part of their creative process. We respect the rights of writers and authors, and believe they should benefit from AI technology. We’re having productive conversations with many creators around the world, including the Authors Guild, and have been working cooperatively to understand and discuss their concerns about AI. We’re optimistic we will continue to find mutually beneficial ways to work together to help people utilize new technology in a rich content ecosystem.”]

Rachel Geman, a partner with Lieff Cabraser and co-counsel for the authors, said that OpenAI’s "decision to copy authors’ works, done without offering any choices or providing any compensation, threatens the role and livelihood of writers as a whole.” She told Ars that "this is in no way a case against technology. This is a case against a corporation to vindicate the important rights of writers.”

  •  Ronnie   ( @Ronnie@lemmy.ca ) 
    link
    fedilink
    English
    531 year ago

    Holy cow the comments here are so alarming. OpenAI isn’t some non profit trying to better humanity. They are a for profit company making money with a product that is ILLEGALY derived from the work of others. Why are y’all so apathetic? Pretty disheartening.

    • Lemmy’s opinion on the topic is often very biased towards the views of tech bros rather than writers/creators, at least that’s what I’ve observed. Tech bros have a boner for LLM AIs. They don’t have anything to lose from the development of these AIs, so they don’t seem to understand the concerns of people who do.

      • Yeah, I was downvoted to hell in a copyright thread for suggesting that my work had worth and that I wasn’t just freely handing it over to the general public. Sounds like a bunch of 12 year olds that have never created a fucking thing in their life except for some artistic skidmarks in their underwear. These kids have a lot to learn about life.

        • Look at any discussion around Sync for Lemmy and you’ll get the same thing. Oh a developer created an app that is a flawless experience so far and looks great, and he wants 20 bucks for it? Burn him at the stake!

          I’m all for FOSS and stuff but people here lean more entitled than they do “free and open”.

        • Personally I write for free, simply because of the joy of it, and what I heard that Google was using people’s Google Drives is free training for their bots, I pulled everything. Not because I want to make a buck, but because Google sure as hell was going to make a buck off of my work without paying me a dime. It’s a little known as principle.

      • Tech bros have a boner for LLM AIs. They don’t have anything to lose from the development of these AIs, so they don’t seem to understand the concerns of people who do.

        On top of this, it’s becoming increasingly clear that many tech bros have never been genuinely moved by a piece of art whether it’s visual or written so they genuinely don’t understand that AI art is devoid of any real emotional impact. AI art just throws together cliches. It reminds me of that shitty AI generated conversation between Plato and Bill Gates when were so many tech bros talking about how “inspiring” it was.

        Don’t get me wrong, I love these AI tools coming out but they’re so over hyped sometimes.

        • The standard for all tech. At the beginning of this I remember them telling me that it was all a new age. Yeah, I’ve been through a few of those now. My favorites were driverless cars and crypto, the same bros told me the same things back then too. Truth is we’ll get a few new really cool things, society will get a bit worse for it, and then we’ll find a new shiny thing.

          LLMs and AI aren’t even new, I studied about them in college. 10 years ago. It’s just that we have faster hardware that can finally support them.

    • I don’t exactly love the outsourcing of human creativity to AI, and personally really hope society continues to value actual human creativity, but the illegality of this stuff simply hasn’t been established.

      What is clear is that directly reproducing copyrighted works is illegal. Additionally, a human taking extensive inspiration from copyrighted works is generally perfectly legal. Some of the key questions are:

      1. Are AI models illegally reproducing copyrighted works?
      2. Does the sheer ease of use and scale of AI make a meaningful difference such that extensive inspiration is illegal?
      3. If AIs being extensively inspired by copyrighted works is in fact legal under current law, should it be? Or should legislation be passed explicitly stating that there’s a material difference due to the drastically different speed and scale of, say, telling an AI to write Winds of Winter compared to an actual human struggling for years and years, and thus AI works need to be treated differently?

      Personally, I’d say that (1) is a maybe, (2) is a no under current law, and therefore (3) is a yes, and I’d love to see legislation passed clarifying how we as a society want to treat AI works. I’m strongly of the opinion that human creativity is something very special and that it should be protected, but I am concerned about a future where that isn’t valued very much in the face of an AI that knows your personal tastes exactly and can simply generate all the “content” you want in an instant.

      • Yup, I remember listening to them talk about how blockchain was going to change the world, and the hype was exactly the same. There are things that will improve, tech will get a bit easier, but the tech will never live up to the hype. They’ll just find something new to chase after when the shine wears off here.

    • I was on board with AI using pre-existing Works to learn off of, but this was before they started hiding their content behind subscription services. Once that happened this became unethical. It’s the difference between having a Pokemon fan game, and trying to publish your own fan mod of Pokemon sword and shield onto the market and expecting Nintendo just be cool with it.

  • The first country to kill off LLMs with draconian interpretations of copyright law will simply be handing the industry to other countries on a golden platter. For this reason, I don’t see the U.S. ruling in any sort of way that would damage the AI industry too much. There is simply too much money involved.

    • If enough countries join in then there will be a barrier to actually making money off of it. Even if you become the leader in AI if your method is just banned in other countries the money won’t be there.

      That being said, it doesn’t seem like it’s going to get very far anyway.

      • Given how we train models (content and math), AIs is not practical to ban/legislate away. While the public applications of AI are for content generation and NLP, as @Rinox alluded to, the military applications are where we are going to see the most focus from the government. As an example, the Lantirn targeting pod uses SVMs to profile aircraft from afar, and it took enormous engineering to get it accurate. Comparable object detection functionality can be obtained with NNs and off-the-shelf GPUs. Countries like China already have “differing philosophies” when it comes to intellectual property rights, so we can remove the largest manufacturing market from the potential list of those who would blanket ban AI. Ditto on any possibility of their military forgoing AI either.

        The real problem here is copyright law, which has extended protections far and above the length of time that is reasonable. Had we terms of say 35 years, we could simply train on older material.

      •  Rinox   ( @Rinox@feddit.it ) 
        link
        fedilink
        English
        41 year ago

        It’s not about making money, not only at least. The other reason why the US, China and other countries are obsessed with AI, LLM, NN, ML is because it may prove decisive for their militaries.

        What is thought by most militaries today is that we are at a turning point of sorts and the next generation of weapons will be powered by AI in some capacity, and the more we go on, the more AI will be involved (target acquisition, reaction, drone and missile guidance, APS, AA, etc)

        • And the very concept of “pirated” textbooks is itself monstrous. Textbooks should be free, in universities, schools, and everywhere. I can see some arguments for copyright maybe possibly being somewhat beneficial for artists (with reasonable limits… for five or maybe even ten years, say, and most definitely no longer than the author’s lifespan, and of course not applicable to companies), but copyrighting learning tools can only be harmful to society.

      • Same principle why Google can index pretty much all books in existence. They were sued over this and won. Same thing will happen here.

        As long as these models are not providing the copyrighted material to their users they should be safe.

        • @Akisamb yes I think you’re right.

          That one was really interesting too. It has put a few more limits on its full text since those days, but I don’t know how much that was a result of the suit.

          The authors from my country tried to have a group lawsuit against Google because within my country, if your books are in public libraries, then you get yearly compensation based on how many copies are in circulation.

          But, Google and America are both a lot more rich and powerful than a handful of authors from New Zealand, so I don’t know what they though they could achieve.

      • Hmm… if that were true then why would they be expecting $150k per book?

        My understanding us that in copyright infringement the liability is the amount of income you have deprived the author of. If you’ve only copied 1 book and not produced derivative works then the loss is the value of the book.

        • @DogMuffins

          This is why it’s such an interesting case. We think of the LLM as one entity, but another way of looking at it is that it’s a lot of iterations.

          The pirated book text files that are doing the rounds with the AI seem to be multiple iterations as well.

          One $150 textbook x 1,000 students = $150k.

          Even with public libraries, they pay volume licenses for ebooks based on how many “copies” can be lended simultaneously per year. It’s not considered as the library owning one copy when digital files are being lent.

          • Sorry mate this is just daft.

            Did you read the article? Most of it is about derivative works.

            LLMs fuel AI tools that “can spit out derivative works: material that is based on, mimics, summarizes, or paraphrases” their works, allegedly turning their works into “engines of” authors’ “own destruction” by harming the book market for them.

            They’re trying to claim that they have been financially harmed by the unauthorised use of their work.

            Even if LLMs are separate iterations you could train multiple LLMs with one copy of the book - the library is not loaning multiple copies simultaneously.

            •  livus   ( @livus@kbin.social ) 
              link
              fedilink
              1
              edit-2
              1 year ago

              @DogMuffins

              Correct me if I’m wrong but I was under the impression they don’t yet have a derivative work that’s close enough to clain plagiarism. Without that, they don’t have a leg to stand on.

              This is the only part I think could potentially hold water:

              Authors should have the right to decide when their works are used to ‘train’ AI. If they choose to opt in, they should be appropriately compensated.”

              As for libraries:

              the library is not loaning multiple copies simultaneously.

              I’m not sure why you’re saying that.

              With ebook platforms the business model completely depends on the concept of multiple copies. This isn’t my opinion, it’s just a fact about how they bill libraries. Random stats.

              Even with physical copies libraries sometimes buy several (and in my country publishers get special compensation based on how many copies are in public libraries per year)

  • If I tried to publish a fan game using the source code of Mega Man Battle network, Capcom would be on my ass if you train your artificial intelligence off of the writing styles of popular authors and they don’t see a lick? They should be on yours

  • This is the best summary I could come up with:


    Martin, Jodi Picoult, and George Saunders joined the Authors Guild in suing OpenAI, alleging that training the company’s large language models (LLMs) used to power AI tools like ChatGPT on pirated versions of their books violates copyright laws and is “systematic theft on a mass scale.”

    They claimed that OpenAI could have trained their LLMs exclusively on works in the public domain or paid authors “a reasonable licensing fee” but chose not to.

    But, the claim adds, so far, Altman and OpenAI—which, claimants allege, “intend to earn billions of dollars” from their LLMs—have “proved unwilling to turn these words into actions.”

    Saunders said that the lawsuit—which is a proposed class action estimated to include tens of thousands of authors, some of multiple works, where OpenAI could owe $150,000 per infringed work—was an “effort to nudge the tech world to make good on its frequent declarations that it is on the side of creativity.”

    On top of this, there are other threats, their complaint said, including that “ChatGPT is being used to generate low-quality ebooks, impersonating authors, and displacing human-authored books.”

    We’re having productive conversations with many creators around the world, including the Authors Guild, and have been working cooperatively to understand and discuss their concerns about AI.


    The original article contains 1,001 words, the summary contains 207 words. Saved 79%. I’m a bot and I’m open source!