Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Japan’s government will not enforce copyrights on data used in AI training

technomancers.ai

Post reply on HN
Print view View original

101–110 of 426 posts

Re: Japan’s government will not enforce copyrights on data used in AI training

#101

Post by nl »

falcolas wrote: post #64
fnordpiglet wrote: post #36

Earlier quoted context omitted.

And if you use it, you’re violating copyright. But you will find no copy of the logo in the model data. The model is way too small to contain its training imagery from an information theoretic point of view.

> But you will find no copy of the logo in the model data. You wont find a copy of a plaintext in a cyphertext. But you can still extract the plaintext from the cyphertext.

If you search "Superman Logo" you find actual copies of the Superman logo which are served from Google's cache.

If you ask a VFX artist to create the "Superman Logo" with Photoshop they'll do an excellent job.

The first one isn't copyright violation because it is fair use. The second maybe if it is redistributed but we don't ban the use of photoshop by artists because they can choose to reproduce copyright things with it.

Re: Japan’s government will not enforce copyrights on data used in AI training

#102

Post by impomura »

failbuffer wrote: post #87
musicale wrote: post #67

Earlier quoted context omitted.

> There's no difference between an art student looking through a museum or archives for ideas and an AI using the material for training. A few notable differences: 1. Scale: a single art student can't view millions of works in a week. 2. Duplication: a single art student's brain can't be cloned or downloaded into another art student's brain. 3. Speed: a single art student cannot draw or paint thousands of images in a…

If an art student was able to gain those abilities, would you then argue they couldn't legally create art?

[flagged]

Re: Japan’s government will not enforce copyrights on data used in AI training

#103

Post by EvanAnderson »

fnordpiglet wrote: post #63
sandworm101 wrote: post #37

Earlier quoted context omitted.

No. Making a single copy for your own use is still a copyright violation. There are exceptions (fair use, nomitive use etc) but just because people are rarely sued for personal copying doesnt equate to that copying being permitted. And trademark issues, such as the other commenter generating the superman logo, are subject to a host of other rules.

Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. There’s no copy of the original, even if it’s able to produce a similar product to the original. That’s the specific thing - the aggregation and the lack of direct reproduction in any form is fundamentally not reproducing or copying t…

> Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff.

Devil’s advocate: That sounds like a derivative work, which would be infringement.

Re: Japan’s government will not enforce copyrights on data used in AI training

#105

Post by paulryanrogers »

fnordpiglet wrote: post #63
sandworm101 wrote: post #37

Earlier quoted context omitted.

No. Making a single copy for your own use is still a copyright violation. There are exceptions (fair use, nomitive use etc) but just because people are rarely sued for personal copying doesnt equate to that copying being permitted. And trademark issues, such as the other commenter generating the superman logo, are subject to a host of other rules.

Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. There’s no copy of the original, even if it’s able to produce a similar product to the original. That’s the specific thing - the aggregation and the lack of direct reproduction in any form is fundamentally not reproducing or copying t…

Depends on the training. Copilot can output training code verbatim. And even if not an exact reproduction, using a small training set could often produce insufficiently transformative work that could still be legally considered a derived work. (IANAL)

Re: Japan’s government will not enforce copyrights on data used in AI training

#107

Post by koboll »

jpollock wrote: post #96
fnordpiglet wrote: post #63

Earlier quoted context omitted.

Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. There’s no copy of the original, even if it’s able to produce a similar product to the original. That’s the specific thing - the aggregation and the lack of direct reproduction in any form is fundamentally not reproducing or copying t…

Copying into RAM during training is making a copy, and can be a copyright violation. https://en.wikipedia.org/wiki/MAI_Systems_Corp._v._Peak_Comp... . However, it seems that there is a later case in the 2nd circuit: https://en.wikipedia.org/wiki/Cartoon_Network,_LP_v._CSC_Hol... .

The difference is that the copy is authorized, unless the work is being pirated.

When an artist displays their work on DeviantArt or Artstation or whatever, they are allowing the general public to load it into memory. It's part of the license agreement they sign when they sign up for these services.

Re: Japan’s government will not enforce copyrights on data used in AI training

#108

Post by tiahura »

EvanAnderson wrote: post #103
fnordpiglet wrote: post #63

Earlier quoted context omitted.

Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. There’s no copy of the original, even if it’s able to produce a similar product to the original. That’s the specific thing - the aggregation and the lack of direct reproduction in any form is fundamentally not reproducing or copying t…

> Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. Devil’s advocate: That sounds like a derivative work, which would be infringement.

converting the original media into a statistical aggregate

Devil's advocate: That sounds transformative, which wouldn't be infringement.

Re: Japan’s government will not enforce copyrights on data used in AI training

#109

Post by nearbuy »

jpollock wrote: post #96
fnordpiglet wrote: post #63

Earlier quoted context omitted.

Training a model isn’t making a copy for your own use, it’s not making a copy at all. It’s converting the original media into a statistical aggregate combined with a lot of other stuff. There’s no copy of the original, even if it’s able to produce a similar product to the original. That’s the specific thing - the aggregation and the lack of direct reproduction in any form is fundamentally not reproducing or copying t…

Copying into RAM during training is making a copy, and can be a copyright violation. https://en.wikipedia.org/wiki/MAI_Systems_Corp._v._Peak_Comp... . However, it seems that there is a later case in the 2nd circuit: https://en.wikipedia.org/wiki/Cartoon_Network,_LP_v._CSC_Hol... .

The sources used for training these AIs are publicly available sources like Common Crawl. If having a copy in RAM is a copyright violation, then there are copyright violations occurring well before any AI ever sees it.

Re: Japan’s government will not enforce copyrights on data used in AI training

#110

Post by slyall »

rocketbop wrote: post #47
scottiebarnes wrote: post #27

Earlier quoted context omitted.

I suppose so. I'm just imagining these "ghost AI artists" who publish catalogues of music using the audible likeness of more prolific artists. I know that you could have always just hired an Eminem impersonator and have them lay down tracks...but this technology lets you achieve speed and scale. At least the Eminem impersonator was a real person. This is just a model learned off an artists voice.

> I'm just imagining these "ghost AI artists" who publish catalogues of music using the audible likeness of more prolific artists. Possibly so, but who is the market for such a catalogue? I don’t see how the artist is going to lose out. I’m reminded, though, if the episode of Mad Men where they want to get the Beatles as the soundtrack to an ad, but on finding out that the Beatles won’t do it they try to get some mus…

A New Zealand political party got into trouble for using a "sound alike" of "Lose Yourself" by Eminem in a political advertisement. They licensed the song from a music library but the court determined the song they licensed was too close to the original.

https://en.wikipedia.org/wiki/Eight_Mile_Style_v_New_Zealand...

Post reply on HN