• @jaybone
    link
    English
    -82 days ago

    I assume they don’t care about the ownership of training data. So they can just train these things on all the pirated data they can get ahold of. Whereas US companies might have a slightly harder time with that.

    • @[email protected]
      link
      fedilink
      English
      62 days ago

      lol did you miss the numerous stories of copyrighted materials and watermarks being spit out by early generative AI?

    • @RubicTopaz
      link
      English
      4
      edit-2
      2 days ago

      Good. Pirate it all if it makes the product FOSS and more efficient.