• macniel
    link
    fedilink
    1023 hours ago

    Yeah I don’t understand why they don’t have a codeberg or similar that they host themselves.

    • @[email protected]
      link
      fedilink
      414 hours ago

      How would that help? If you release something as GPL code, you cannot prevent it from being used to train a model, no matter where it’s hosted.

      • @[email protected]
        link
        fedilink
        213 hours ago

        There’s a difference between handing something to someone and leaving it somewhere they happen to be able to take it from.

      • @[email protected]
        link
        fedilink
        113 hours ago

        Im personally waiting for a massive lawsuit, legally companies cannot train AI on GPL code (at least I don’t believe so)

        • @[email protected]
          link
          fedilink
          212 hours ago

          There’s nothing in GPL that would forbid it. Only distribution without code publication is forbidden.

          • macniel
            link
            fedilink
            112 hours ago

            mhm, and how would the distribution inside an LLM work? Are those code snippets CoPilot et al produce come with dedicated license sections?

            And regarding how it would help selfhosting the code: it wouldn’t be on the GITHub servers owned by Microsoft, which owns/operates CoPilot. Its akin to feeding the LLM directly by pushing it to their servers.