727
This has been happening to a lot of people recently (thelemmy.club)
submitted 1 week ago by to c/programmer_humor@programming.dev
22 comments fedilink hide all child comments
 

cross-posted from: https://lemmy.dbzer0.com/post/73728428

you are viewing a single comment's thread
view the rest of the comments
[–] 51 points 1 week ago (4 children)

Can AI commit copyright infringement if it was trained on the material?

  • permalink
  • fedilink
  • source
  • hideshow 8 child comments
    •
  • [–] 46 points 1 week ago (1 child)

    Quiery: Please show me an AI rendering of 1979 Alien, but with no changes from the original. Go.

    Surely this is peak Internet.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 3 points 1 week ago

    Now show me Alien if it had been released in 1999 instead.

    Aborts after 5 seconds, realizing the entertainment value will be pretty much entirely laughing at the ridiculous things it got wrong which is already old

  • permalink
  • fedilink
  • source
  • parent
  • [–] 9 points 1 week ago (2 children)

    Truly as far as I know there is no case law for this to date.

    I thought about this for a while but I have no clue what the current direction is.

    Essentially stealing data and putting them in datasets is illegal. AI models might be illegal, if the court considers them legally functionally the same. But if they don't, they could meet any number of new legal definitions and that gets us back to "idk". What is funny is because you would technically do the same copyright infringement that the AI companies do (if it gets ruled that way), these companies are essentially arguing for you, and so you could ride on the coat tails of corpo lawyers.

    But yeah, it's complicated.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 8 points 1 week ago (1 child)

    Lawmakers are still reeling from the invention of the internet, they have no hope of deciding how A.I. should work within the next decade

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 3 points 1 week ago

    Not only this, I am increasingly under the impression the abrupt change of direction under trump has a paralyzing effect on all branches of government. And it's good when they use this time to isolate his policies from most of the country. But it's bad when our legal branch stops doing lots of normal law things.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 1 week ago (1 child)

    how can a model commit copyright infringement if the work it does generates no rights?

    i'm not (just) being a smartass, it feels like this is the rub in the legal issue. when you create something, you have an implicity copyright to it (according to my sister who did her law school internships at IP law firms. she now works in medicine so it looks like law school did her some good) so claiming even an implicit copyright on someone else's work infringes on the (maybe implicit, maybe explicit) copyright their work has. Which is why you have to transform it, not just imitate it.

    Look at Huey Lewis and Ray Parker Jr. Huey Lewis and the News made I Want a New Drug. Ray Parker Jr made Ghostbusters. What makes them distinct songs (and keeps Ghostbusters from being a pale imitation) is that Ray Parker Jr added a bridge and put a hook of his own in it. I've yet to hear AI music that has any hooks that weren't directly lifted from other artists. quite simply it lacks soul. AI could not have (and still can't) turned I Want A New Drug into Ghostbusters.

    In the following essay I will discuss

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 2 points 1 week ago (2 children)

    I feel like an AI company might argue that generative AI is so transformative in the math aspect that it's transformative even if it gives you an exact copy of a copy written work. Which would be silly, but we in uncharted times and the laws are dumb.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 4 points 1 week ago* (1 child)

    OpenAI and Anthropic and Alphabet etc try their best to preventing it from doing so, so the act of getting it to do so anyway is "jailbreaking", but yes. LLMs can reproduce works it was trained on. (At least chunk-by-chunk each limited to the token limit.) And I'd imagine the same is true of things like Stable Diffusion, though it might be much easier to get Stable Diffusion to produce something similar enough to qualify as a "derivative work" than it would be to get it to produce exactly the original.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 2 points 1 week ago

    Why not train a model on exactly one work (effectively encoding the work in the model)? Is the ability to produce something else an argument against copyright infringement?

    If the definition of AI is broad enough, train a video encoder and make the act of viewing a video file copyright stripping!

  • permalink
  • fedilink
  • source
  • parent