Japan’s government will not enforce copyrights on data used in AI training • Hacker News Board

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

#341

Post by jtode » Thu, Jun 01, 2023, 10:27 AM UTC

LadyCailin wrote: post #84

jtode wrote: post #76

Earlier quoted context omitted.

Big difference between the art student producing work and getting the credit vs you taking the art student's work and taking the credit for it. The AI is not a human, but what you are doing is the same thing, if you claim the output as your work because you wrote the prompt.

I couldn’t help but notice you didn’t credit any web browser in this comment. And rightfully so. Software doesn’t need or care about being credited. Well, usually. Sent from my iPhone.

Lol sure buddy, that browser is ai powered and I didn't give the browser an address, I have it a prompt describing the type of site that I would like it to generate for me.

Edit: and the browser have the same page to you and me. I bet you look at everything and think about how you can make lazy money with it.

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

#342

Post by dooraven » Thu, Jun 01, 2023, 10:31 AM UTC

nologic01 wrote: post #286

This article is an example of emerging AI-bro tactics that completely mirrors crypto-bro tactics: they pick any piece of news and reinterpret it to fit an agenda. While the article is in English, the link to source is in Japanese. The only external source I found suggests the discussion is about promoting open data and open science from research institutions [1] [1] https://asianews.network/japan-to-promote-use-of-ge…

The Japanese article does explicitly state it if you run it through a translator, and also this is from May 11

> Additionally, the group raised other issues that Article 30-4 of the Copyright Law, which permits the use of a copyrighted work for machine learning, does not include procedures for gaining permission in advance from copyright holders. The article permits the use of copyrighted material such as text and images to train AI, regardless of whether the model is for commercial use. Under the current law, it is legal to train AI with copyrighted material even if the data was obtained illegally. The article contains a provision stating that such material cannot be used if it would “unreasonably prejudice the interests of the copyright owner,” but there are only limited examples provided to describe the “unreasonable prejudice.”

https://www.lexology.com/library/detail.aspx?g=d8b4ba7d-a764...

Right now, Japanese copyright doesn't apply to training models. Could change in the future but the article isn't inaccurate.

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

#343

Post by DennisP » Thu, Jun 01, 2023, 10:34 AM UTC

kmod wrote: post #114

fnordpiglet wrote: post #19

I think this should generally be true. The aggregation performed by model training is highly lossy and the model itself is a derived work at worst and is certainly fair use. It may produce stuff that violates copyright, but the way you use or distribute the product of the model that can violate copyright. Making it write code that’s a clone of copyright code or making it make pictures with copy right imagery in it or…

This is the "guns don't kill people, people do" argument. Not saying that that proves things one way or another, just that assigning responsibility is not really a cut and dry question and many people think that it's important to look prior to the final interaction IANAL but I believe in the US tools that are designed to circumvent copyright are illegal, which makes sense to me inasmuch as one believes that copyright…

Photocopy machines are great at making copies of copyrighted material, and are completely legal in the US. The entire internet routinely makes copies every time you visit a web page.

What's illegal in the US is selling tools for breaking DRM: https://www.androidpolice.com/2018/10/26/us-copyright-office...

From a quick google, open source tools for breaking DRM are legal, and so is breaking DRM for personal use: https://www.google.com/search?q=are+drm+defeating+tools+ille...

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

#344

Post by pxoe » Thu, Jun 01, 2023, 10:38 AM UTC

jupp0r wrote: post #289

pxoe wrote: post #287

Earlier quoted context omitted.

Some licenses, like CC, have variants that prohibit production of derivatives, or prohibit commercialization, or require licenses or same requirements to be preserved in derivative works. Sweet to see people "warm up" to say stuff like 'well it's just a derived work'. Can we get tech to actually respect the licenses of used works next? It's something that's been asked for all along. Without just going, 'well, it's al…

Did you just argue against fair use in general? Everything you wrote applies to it as well. Copyright has limits, it's not like right holders get to determine what those are. It's a balance of rights of creators and users.

is the use really all that fair, when thousands, if not millions, of works and artists get their works repurposed into services with "subscriptions" and "usage tokens" and other kinds of monetization, while directly competing with and against those very artists? or will it take until some markets would get completely destroyed and overtaken while people get displaced, for people to wake up and go 'wait, was that really "fair"? what happened?'

and no, im not really arguing against it. fair use can be great. but shit like that, it's really pushing it to its limits on scope and scale of use and commercialization

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

#345

Post by jtode » Thu, Jun 01, 2023, 10:42 AM UTC

Aachen wrote: post #312

walthamstow wrote: post #310

Earlier quoted context omitted.

Not just crypto bros, this kind of thing is rife in politics too. Brexit is full of it. People pick one article about one minor thing in one niche area of the economy and use it to 'prove' their entire agenda.

Not just politics. I remember this being a realisation as a teenager, noticing that if you bring five reasons, the person you're talking to will refute a random one in a funny way and now the audience will decide you were wrong. Danny was the person who was absolutely the best at this. I should have written one of them down, as I can't even reproduce it but he'd use some logical fallacy to make his case which is, for…

There's a name for this: The Gish Gallop. It was not invented but was the specialty of one Duane T. Gish, a creationist who specialized in "winning" debates with unsuspecting academics for a while until the community decided to stop playing chess with a pigeon.

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

#346

Post by quicklime » Thu, Jun 01, 2023, 10:45 AM UTC

nl wrote: post #89

fnordpiglet wrote: post #19

I think this should generally be true. The aggregation performed by model training is highly lossy and the model itself is a derived work at worst and is certainly fair use. It may produce stuff that violates copyright, but the way you use or distribute the product of the model that can violate copyright. Making it write code that’s a clone of copyright code or making it make pictures with copy right imagery in it or…

I strongly agree with this. There's a distinction between "learning from" and "copying". "Learning from" is a transformative process that distills from the observation. This distillation can be as simple as indexing for a search engine, or as complex as a deep neural network. Simply because a neural network can create something that is a copyright violation doesn't mean the training process itself it. A human can see…

I strongly disagree with this. We shouldn't create new laws for new technology by making analogies to what's allowed under old laws designed for old technology. If we did, we would never have come up with copyright in the first place.

600 years ago, people were allowed to hand-copy entire books, so they should be able to do it with a printing press right? It's "just a tool"!

The correct way to think about this is to recognize that society needs people to create training data as well as people to train models. If we don't reward the people who create training data, we disincentivize them from doing so, and we'll end up in a world where we don't have enough of it.

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

#347

Post by nologic01 » Thu, Jun 01, 2023, 10:48 AM UTC

dooraven wrote: post #342

nologic01 wrote: post #286

This article is an example of emerging AI-bro tactics that completely mirrors crypto-bro tactics: they pick any piece of news and reinterpret it to fit an agenda. While the article is in English, the link to source is in Japanese. The only external source I found suggests the discussion is about promoting open data and open science from research institutions [1] [1] https://asianews.network/japan-to-promote-use-of-ge…

The Japanese article does explicitly state it if you run it through a translator, and also this is from May 11 > Additionally, the group raised other issues that Article 30-4 of the Copyright Law, which permits the use of a copyrighted work for machine learning, does not include procedures for gaining permission in advance from copyright holders. The article permits the use of copyrighted material such as text and im…

How does the "Japan’s government will not enforce copyrights" message of the article square with your own source (thanx btw :-)

> It will be suggested that the current Japanese Copyright Law will be reviewed based on the upcoming technologies and be revised to adjust to correctly protect the right holders and navigate future users of the AI and new technologies like AI

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

#348

Post by pravus » Thu, Jun 01, 2023, 10:51 AM UTC

Mordisquitos wrote: post #295

inkblotuniverse wrote: post #293

Earlier quoted context omitted.

If I read a lot of fantasy books as a kid, then start writing my own fantasy book, should I have to pay royalties to the authors of the books I read?

Does your ability to write fantasy books absolutely depend on having read those fantasy books as a kid? Was gaining the ability to write your own fantasy books and profit from them your only motivation to read those fantasy books? After gaining the ability to write fantasy books thanks to having read them, can you now produce fantasy books at a qualitatively different speed, scale, and conditions than any of the auth…

How much does the world owe to Gilgamesh and Homer?

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

#349

Post by cthalupa » Thu, Jun 01, 2023, 10:57 AM UTC

nologic01 wrote: post #347

dooraven wrote: post #342

Earlier quoted context omitted.

The Japanese article does explicitly state it if you run it through a translator, and also this is from May 11 > Additionally, the group raised other issues that Article 30-4 of the Copyright Law, which permits the use of a copyrighted work for machine learning, does not include procedures for gaining permission in advance from copyright holders. The article permits the use of copyrighted material such as text and im…

How does the "Japan’s government will not enforce copyrights" message of the article square with your own source (thanx btw :-) > It will be suggested that the current Japanese Copyright Law will be reviewed based on the upcoming technologies and be revised to adjust to correctly protect the right holders and navigate future users of the AI and new technologies like AI

My reading is that people are suggesting that the copyright law be reviewed and changed, not that lawmakers are suggesting that they will change it. Lots of people are suggesting that the US and other western countries change the law as well, but we have yet to see much come from it.

The general message seems to be that 'Under current laws Japan's government can't enforce copyright on training data', and I don't believe the line you're quoting changes that message in any significant way at current.

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

#350

Post by dooraven » Thu, Jun 01, 2023, 11:01 AM UTC

nologic01 wrote: post #347

dooraven wrote: post #342

Earlier quoted context omitted.

The Japanese article does explicitly state it if you run it through a translator, and also this is from May 11 > Additionally, the group raised other issues that Article 30-4 of the Copyright Law, which permits the use of a copyrighted work for machine learning, does not include procedures for gaining permission in advance from copyright holders. The article permits the use of copyrighted material such as text and im…

How does the "Japan’s government will not enforce copyrights" message of the article square with your own source (thanx btw :-) > It will be suggested that the current Japanese Copyright Law will be reviewed based on the upcoming technologies and be revised to adjust to correctly protect the right holders and navigate future users of the AI and new technologies like AI

Well they have to follow the current law right? Can't enforce copyright on something that doesn't apply, so currently saying they won't enforce copyright is an accurate statement.

I don't think it will hold up in the long term though and the Copyright Act will get changed but right now it doesn't apply to training models.