• John@lemmy.ca
    link
    fedilink
    English
    arrow-up
    0
    ·
    2 days ago

    I guess all that bad code people have been prompting AI to write is public domain, no matter what license they assign to it. Licenses are all based on copyright.

    • NaibofTabr@infosec.pub
      link
      fedilink
      English
      arrow-up
      0
      ·
      2 days ago

      Heh, this would be fantastic and I think technically in line with the GNU GPL - all code produced from GPL code must also be licensed as GPL. Therefore the output of any model that trained on any GPL code would also be GPL.

      Open source all the things!

      • John@lemmy.ca
        link
        fedilink
        English
        arrow-up
        0
        ·
        2 days ago

        The GPL is a license based on copyright. If AI output can’t be copyrighted, then it can’t be licensed, it becomes public domain, which means you can make a derivative, then copyright that under a commercial license.

        With no.copyright, stuff gets weird fast.

        • Rekall Incorporated@piefed.social
          link
          fedilink
          English
          arrow-up
          0
          ·
          1 day ago

          Wait, so if you include LLM-derived source code into a GPL project, it loses it’s access to copyright?

          This doesn’t make sense…

          And one would thing the same would have to be true of proprietary software.

        • NaibofTabr@infosec.pub
          link
          fedilink
          English
          arrow-up
          0
          ·
          1 day ago

          No no, see the GNU GPL is copyleft:

          The licenses in the GPL series are all copyleft licenses, which means that any derivative work must be distributed under the same or equivalent license terms.

          https://en.wikipedia.org/wiki/GNU_General_Public_License

          So if (stressing the if) output from an AI that was trained on GPL code is considered a derivative work, then it must also be licensed as GNU GPL. That makes it open source, but not unlicensed.

          GNU GPL is intentionally insidious this way, it prevents corporate profiteering from GPL projects because any derivative work must use the same license.

          The question is whether a court decision would uphold that AI generated code based on GPL code training counts as a derivative work. This decision regarding generated art seems like it might set a precedent for that.

          • John@lemmy.ca
            link
            fedilink
            English
            arrow-up
            0
            ·
            22 hours ago

            The GPL is based on rights from copyRIGHT law. It’s the author’s copyright that allows them to determine how to license it, and the GPL is one of many licenses.

            So if something is not able to be copyrighted, then it’s not possible to put a copyright license like the GPL on it. The work is in the public domain, no license at all, different rules about derivative works.

            If machines generated works (like code) cannot be copyrighted, then they can’t be licensed. Is the output of an LLM trained on GPL a derivative work, but can’t be copyrighted?

            It’s a crazy decision, it’s going to do a lot of weird stuff, at least in the US. It might invalidate a bunch of the IP treaties that the US negotiated with much of the world since 2000.

  • supersquirrel@sopuli.xyz
    link
    fedilink
    English
    arrow-up
    0
    ·
    2 days ago

    This is racist against computers, my AI is sentient and there is solid legal precedent for owning sentient beings in the US!!!