5 ms·

Because there is a lot of money and fear involved I agree that there will be a lot of litigation, but the outcomes will be mixed at best for the copyright holde

Because there is a lot of money and fear involved I agree that there will be a lot of litigation, but the outcomes will be mixed at best for the copyright holders. Lots of people have memorized lyrics to copyrighted songs but no one is suing them for the memorization. They are only suing when the memorizer is trying to make money off of the song without permission of the copyright holder. The burden of proof is on the copyright holder to prove that the builder of the model was intending to distribute IP that they did not own.

mikkergp 4y ago
> They are only suing when the memorizer is trying to make money off of the song without permission of the copyright holder. Granted everything I know about copyright law I learned in a few audio tapes in my car, but I think the question I have is not, can you sue AI for absorbing information, but is AI unique enough that you couldn't find byte strings or patterns that match source material too closely. See the recent twitter thread on CSParse for example: https://twitter.com/docsparse/status/1581461734665367554 https://twitter.com/docsparse/status/1581461734665367554
- gitfan86 4y ago
  It is just weights, there is no string of bytes
  - mikkergp 4y ago
    I’m talking about outputs not the model itself, the GitHub copilot example has lots of shared bytes
  - lbotos 4y ago
    You are correct that the "model" may be a separate piece of work with it's own copyright but that does not mean that the model does not produce copyrighted works.
pdntspa 4y ago
A lot of people have sampled literally everything in music and it is very very uncommon to get sued for creating something substantially different. I fail to see how AI is any different. People need to realize that IP enforcement is a very leaky sieve at best, and that the very concept of intellectual property is an affront to the nature of information and ideas, particularly in a digital realm where bits are free to be copied at will The sooner we come to this realization, and those that stand to profit from IP enforcement bullshit take their monetization elsewhere, the better off we will be as a species.
- capableweb 4y ago
  A lot of established artists (or upcoming artists) don't just randomly sample stuff, release it and call it a day. Usually songs go through a "sample clearing" pass where all the samples are "cleared", meaning basically licensed for use. Usually you don't do that until right before releasing it (it's not part of the conception of the song, to make sure you're using cleared samples), but most popular released music has cleared samples before releasing the songs. If you're not familiar with the process, you can read about it here: https://medium.com/the-courtroom/the-art-of-clearing-a-sample-deciding-if-its-worth-it-and-how-to-actually-do-e26fa56ad090 https://medium.com/the-courtroom/the-art-of-clearing-a-sampl...
  - pdntspa 4y ago
    I don't think that most of Beatport's or Bandcamp's catalogue has gone through that clearance process, and I'd wager a large part of Spotify (via tunecore/etc) hasn't either Nobody has ever given me any crap for the blatant sampling I do from old movies and video games, though I am a nobody and they are always a small (if focal) part of a much larger and original work
    - capableweb 4y ago
      They probably haven't, yet ;) But if a song (yours or other people's songs on Beatport/Bandcamp/Spotify) becomes popular enough and it has samples you haven't cleared, expect your door to be knocked on by some of the right holders who are gonna ask you to pay up. There are plenty of cases of unknown people making songs that become popular overnight and then get stuck in long litigations regarding sample usage. Obviously they're not gonna go after music makers who barely have any listeners.
phaedrus 4y ago
It's premature to claim these "AI" models are equivalent to a human memorizing lyrics to copyrighted songs. They're much closer to XOR'ing copyrighted works so that the encoded form is not casually recognizable. I don't have sources close to hand, but I seem to recall during the 00's reading about such schemes being regularly proposed and also regularly struck down, legally. From a high level both things consist of putting copyrighted bits in a digital blender, producing a mixture unlike the originals but which, if you prod it the right way, can be induced to reproduce some parts of the (copyrighted) input. I'm not sure calling it AI makes it (legally) different. What is the argument here - that previous schemes just didn't blend the bits enough?
- gitfan86 4y ago
  It isn't premature at all. Stable Diffusion took 100GB of already compressed images and turned them into a 2GB model. The XOR of those 100GM images could never fit into 2GB.
  - Karliss 4y ago
    All this argument proves is that model it doesn't contain lossless full size byte perfect copies of all the image. And that's assuming all those 100GB of are images are unique with no common data. Copying 2% of those images or maybe 8% at 50% width would still be a problem.Just the difference between high and low quality JPG compression can easily make 10x size difference. You can start to see visible artifacts at that compression levels, but I wouldn't try to convince a movie studio that badly compressed version of their movie doesn't count as copyright infringement.
    - gitfan86 4y ago
      But that is exactly how copyright law works. If you copy the first 10000 bytes of a steven king book and compress it with zip and distribute it, there is a good chance you would lose the lawsuit if sued. If on the other hand you copy the first, middle, and last byte of a Steven King book and share those three bytes with someone you would win a copyright lawsuit, because those three bytes are not unique enough to be copyrighted.

mikkergp 4y ago

> They are only suing when the memorizer is trying to make money off of the song without permission of the copyright holder. Granted everything I know about copyright law I learned in a few audio tapes in my car, but I think the question I have is not, can you sue AI for absorbing information, but is AI unique enough that you couldn't find byte strings or patterns that match source material too closely. See the recent twitter thread on CSParse for example: https://twitter.com/docsparse/status/1581461734665367554 https://twitter.com/docsparse/status/1581461734665367554

gitfan86 4y ago
It is just weights, there is no string of bytes
- mikkergp 4y ago
  I’m talking about outputs not the model itself, the GitHub copilot example has lots of shared bytes
- lbotos 4y ago
  You are correct that the "model" may be a separate piece of work with it's own copyright but that does not mean that the model does not produce copyrighted works.

pdntspa 4y ago

A lot of people have sampled literally everything in music and it is very very uncommon to get sued for creating something substantially different. I fail to see how AI is any different. People need to realize that IP enforcement is a very leaky sieve at best, and that the very concept of intellectual property is an affront to the nature of information and ideas, particularly in a digital realm where bits are free to be copied at will The sooner we come to this realization, and those that stand to profit from IP enforcement bullshit take their monetization elsewhere, the better off we will be as a species.

capableweb 4y ago
A lot of established artists (or upcoming artists) don't just randomly sample stuff, release it and call it a day. Usually songs go through a "sample clearing" pass where all the samples are "cleared", meaning basically licensed for use. Usually you don't do that until right before releasing it (it's not part of the conception of the song, to make sure you're using cleared samples), but most popular released music has cleared samples before releasing the songs. If you're not familiar with the process, you can read about it here: https://medium.com/the-courtroom/the-art-of-clearing-a-sample-deciding-if-its-worth-it-and-how-to-actually-do-e26fa56ad090 https://medium.com/the-courtroom/the-art-of-clearing-a-sampl...
- pdntspa 4y ago
  I don't think that most of Beatport's or Bandcamp's catalogue has gone through that clearance process, and I'd wager a large part of Spotify (via tunecore/etc) hasn't either Nobody has ever given me any crap for the blatant sampling I do from old movies and video games, though I am a nobody and they are always a small (if focal) part of a much larger and original work
  - capableweb 4y ago
    They probably haven't, yet ;) But if a song (yours or other people's songs on Beatport/Bandcamp/Spotify) becomes popular enough and it has samples you haven't cleared, expect your door to be knocked on by some of the right holders who are gonna ask you to pay up. There are plenty of cases of unknown people making songs that become popular overnight and then get stuck in long litigations regarding sample usage. Obviously they're not gonna go after music makers who barely have any listeners.

phaedrus 4y ago

It's premature to claim these "AI" models are equivalent to a human memorizing lyrics to copyrighted songs. They're much closer to XOR'ing copyrighted works so that the encoded form is not casually recognizable. I don't have sources close to hand, but I seem to recall during the 00's reading about such schemes being regularly proposed and also regularly struck down, legally. From a high level both things consist of putting copyrighted bits in a digital blender, producing a mixture unlike the originals but which, if you prod it the right way, can be induced to reproduce some parts of the (copyrighted) input. I'm not sure calling it AI makes it (legally) different. What is the argument here - that previous schemes just didn't blend the bits enough?

gitfan86 4y ago
It isn't premature at all. Stable Diffusion took 100GB of already compressed images and turned them into a 2GB model. The XOR of those 100GM images could never fit into 2GB.
- Karliss 4y ago
  All this argument proves is that model it doesn't contain lossless full size byte perfect copies of all the image. And that's assuming all those 100GB of are images are unique with no common data. Copying 2% of those images or maybe 8% at 50% width would still be a problem.Just the difference between high and low quality JPG compression can easily make 10x size difference. You can start to see visible artifacts at that compression levels, but I wouldn't try to convince a movie studio that badly compressed version of their movie doesn't count as copyright infringement.
  - gitfan86 4y ago
    But that is exactly how copyright law works. If you copy the first 10000 bytes of a steven king book and compress it with zip and distribute it, there is a good chance you would lose the lawsuit if sued. If on the other hand you copy the first, middle, and last byte of a Steven King book and share those three bytes with someone you would win a copyright lawsuit, because those three bytes are not unique enough to be copyrighted.