Comment on Japan determines copyright doesn't apply to LLM/ML training data

• LWD@lemm.ee ⁨10⁩ ⁨months⁩ ago

  Painting and selling an exact copy of a recent work, such as Banksy, is a crime.

  … however making an exact copy of Banksy for personal use, or to learn, or to teach other people, or copying the style… that’s all perfectly legal.

  And that was the bait and switch of OpenAI! They sold themselves as being a non-profit simply doing research, for which it would be perfectly legal to consume and reproduce large quantities of data… And then, once they had the data, they started selling access to it.

  I would say that that alone, along with the fact that they function as gatekeepers to the technology (One does not simply purchase the model from OpenAI, after all) they are hardly free of culpability… But it definitely depends on the person trying to use their black box too.

  source

  • abhibeckert@lemmy.world ⁨10⁩ ⁨months⁩ ago

    Huh? What does being non profit have to do with it? Private companies are allowed to learn from copyrighted work. Microsoft and Apple, for example, look at each other’s software and copy ideas (not code, just ideas) all the time. The fact Linux is non-profit doesn’t give them any additional rights.

    source
  • iegod@lemm.ee ⁨10⁩ ⁨months⁩ ago

    They’re not gatekeeping llms though, there are publicly available models and data sets.

    source

    • LWD@lemm.ee ⁨10⁩ ⁨months⁩ ago

      If it’s publicly available, why didn’t Microsoft just download and use it rather than paying them for a partnership?

      IIRC they only open-sourced some old stuff.

      source

      • iegod@lemm.ee ⁨10⁩ ⁨months⁩ ago

        Stability diffusion is open source. You can run local instances with provided and free training sets to query against and generate your own outputs.

        stability.ai

        source
• silverbax@lemmy.world ⁨10⁩ ⁨months⁩ ago

  Thanks for your response. I realize I muddied the waters on my question by mentioning exact copies.

  My real question is based on the ‘everything is a remix’ idea. I can create a work ‘in the style of Banksy’ and sell it. The US copyright and trademark laws state that a work only has to be 10% differentiated from the original in order to be legal to use, so creating a piece of work that ‘looks like Banky, but not done by Banksy’ is legal.

  So since most AI does not create exact copies, this is where I find the licensing argument possibly weak. I really haven’t seen AI like MidJourney creating exact replicas of works - but admittedly, I am not following every single piece of art created on Midjourney, or Stable Diffusion, or DALL-E, or any of the other platforms, and I’m not an expert in the trademarking laws to the extent I can answer these questions.

  source

  • abhibeckert@lemmy.world ⁨10⁩ ⁨months⁩ ago

    The US copyright and trademark laws state that a work only has to be 10% differentiated from the original in order to be legal to use

    I’m pretty sure the law doesn’t say that.

    source
  • Silentiea@lemm.ee ⁨10⁩ ⁨months⁩ ago

    I really haven’t seen AI like MidJourney creating exact replicas of works

    I don’t have a source to cite, but I did read an article that showed a bad faith actor deliberately trying to use ai to copy images directly, and while the results weren’t exact replicas, they were reasonable facsimiles of the original, to the extent that if a human has created it without ai, it would have been blatant copyright infringement, despite not being quite identical.

    I wish I had the examples on hand to show, but it was months ago, and unfortunately I have not the skills nor time to retrieve it.

    source
• tabular@lemmy.world ⁨10⁩ ⁨months⁩ ago

  To be at fault the user would have to know the AI creation they distributed commits copyright infringement. How can you tell? Is everyone doing months of research into every creative works to make sure it’s not somewhat like someone else’s?

  Even if you had an AI trained on only public domain assets you could still end up putting in the words that generate something copyrighted, and not know.

  Companies created a random copyright infringement tool and users randomly infringe copyright.

  source

  • red@sopuli.xyz ⁨10⁩ ⁨months⁩ ago

    The same way you can tell if you repainted a Banksy yourself. If you don’t realize, and monetize, then you are liable for a copyright lawsuit regardless of the wat you created the piece in question.

    source

    • tabular@lemmy.world ⁨10⁩ ⁨months⁩ ago

      You may recognize a Banksy but to another it’s like I said you aught to know your work is like one from Coinsey: who?

      This is exasperated when people can create creative works via AK, having even less knowledge about your peers who know how to DIY. A potentially life-ruining lawsuit is a bad system to find out you can’t monetize something.

      source

      • red@sopuli.xyz ⁨10⁩ ⁨months⁩ ago

        If only there was some way to find out prior to selling stuff as if you made it. If only. Darn it!

        source

        • -> View More Comments
• Mango@lemmy.world ⁨10⁩ ⁨months⁩ ago

  Your example is a dude who paints unsolicited on other people’s property. What kind of copyright does a ghost have?

  source

  • Silentiea@lemm.ee ⁨10⁩ ⁨months⁩ ago

    A surprising amount, though it would potentially be quite difficult to prove.

    source

    • Mango@lemmy.world ⁨10⁩ ⁨months⁩ ago

      I should paint some shit on your house and then sue you for displaying it.

      source