• @besbin@lemmy.ml
    link
    fedilink
    211 months ago

    Whatever already existed won’t be thrown away regardless of the ruling. It’s like throwing all the gold already dug up just because it was done by slave labor. The law and legal actions are mostly just a moat around the pile of gold already dug up. Sure AI companies will have to pay more for the new data from other sources. However that would be peanut compared to how much they will have to pay starting from zero.

    • @Ross_audio@lemmy.world
      link
      fedilink
      111 months ago

      If every time what already exists gets used there’s a risk of a massive fine or court case they’ll throw it away.

      The game now is to delay the legal process long enough until they’ve built the replacement.

      Then they can afford to throw the, essentially faulty, model away.

        • @Ross_audio@lemmy.world
          link
          fedilink
          111 months ago

          It’s clear from the output that it breaks copyright.

          We don’t have to look inside the black box to demand to see the input which caused that output.

          To be clear a machine is not responsible for itself. This machine was trained to break copyright.

          • Generally if someone is clearly in breach of copyright the rights holder will apply to a court to issue an injunction to order that company to cease their activities until a case can be resolved.

            Given that has not happened, it seems that from a court’s perspective, it’s not a clear breach of copyright.

            • @Ross_audio@lemmy.world
              link
              fedilink
              111 months ago

              The rights holder first considers the size of the payout vs. the cost of legal fees.

              Just because they haven’t been sued directly for this doesn’t make it infringement.