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

#371

Post by bastawhiz » Thu, Jun 01, 2023, 3:23 PM UTC

carom wrote: post #210

bastawhiz wrote: post #123

Earlier quoted context omitted.

I don't fundamentally disagree with you, but what you are saying doesn't hold water. > a copy is made and reproduced numerous times when training. Casually browsing the web creates millions of copies of what are likely the same images and text that models are trained on. Computers cannot move information, they can only copy it and delete the original. Splitting hairs over the semantics of what it means to "copy" isn'…

There is quite a bit of precedent for "making copies of digital things is copyright infringement". Look at lawsuits from the Napster era. [1] What makes the use improper? Licenses. Terms of service. Mostly licenses though. For example, all the images on Flickr that were uploaded under Creative Commons licenses (e.g. non-commercial) have now been used in a commercial capacity by a company to create and sell a product.…

> to create and sell a product.

This is not model training.

> Copilot is a derivative work of that code, the license terms of that code (e.g. GPL, non-commercial) should extend to the new function that was derived from it.

But the very act of training copilot is not problematic. And in fact, if GitHub never did anything with Copilot, the physical act of training the model is not problematic at all. And that's what at issue here. How Copilot is used is orthogonal to the article.

> Sure, I can make a shirt with Spider Man on it and give it to my brother, but if a company were to use what I made or I tried to sell it, I would expect a cease and desist from Disney.

Yes. And training the model isn't the part where you sell it. It's the part where you make it.

> Training the model may very well be a copyright issue. The images have been copied, they are being used.

What do you think "being used" means here? If I work for a company and download a bunch of text and save it to a flash drive, have I violated copyright? Of course not. If I put that data in a spreadsheet, is it copyright infringement? Of course not. If I use Excel formulas on that text is it infringement? Still no.

And so how can you claim in any way that the creation of a model is anything more than aggregating freely available information?

I don't disagree with you about the use of a model. But training the model is just taking some information and running code against it. That's what's important here.

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

#372

Post by WhereIsTheTruth » Thu, Jun 01, 2023, 3:46 PM UTC

That's understandable, but we must be careful not to scare content creators, they rely on the little their art provide them, I feel like we should be more social in a world driven by generative AI, specially when the AI companies are driven by profit

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

#373

Post by leni536 » Thu, Jun 01, 2023, 4:08 PM UTC

adam_arthur wrote: post #182

lamontcg wrote: post #132

Earlier quoted context omitted.

> There's a distinction between "learning from" and "copying". Neural nets can memorize their training data. Generally that isn't what you want, and you strive to eliminate it. However, it could instead be encouraged to happen if someone wanted to exploit this law in order to abuse copyrights.

Humans can memorize their training data too... aka see something and then produce a copy (code, drawing, music etc). The principles underlying how LLMs and humans learn isn't really that different... just different levels of loss/fuzziness.

And when humans do that they may also infringe on copyright.

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

#374

Post by jpollock » Thu, Jun 01, 2023, 4:10 PM UTC

koboll wrote: post #107

jpollock wrote: post #96

Earlier quoted context omitted.

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.

The copy isn't authorized, the copy is allowed under Fair Use. There's a huge difference between the two.

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

#375

Post by jpollock » Thu, Jun 01, 2023, 4:16 PM UTC

fnordpiglet wrote: post #129

jpollock wrote: post #96

Earlier quoted context omitted.

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... .

How do search engines exist? The internet archive? Caching of image results? Web browser caches? CDNs?

Copies made by search engines don't need authorization, and can be unauthorized copies. Search engines are allowed to make copies under Fair Use since they are transformative - see Authors Guild, Inc. v. Google, Inc.

There hasn't been an explicit decision for ML training, but everyone's assuming that Authors Guild v Google applies.

https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,.....

CDNs operate under the control of the copyright owner, so they would be authorized.

Web browser caches are under the control of the recipient who has authorization to make a copy.

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

#376

Post by jpollock » Thu, Jun 01, 2023, 4:27 PM UTC

harshreality wrote: post #134

jpollock wrote: post #96

Earlier quoted context omitted.

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... .

MAI v. Peak was obviously wrong. It would mean whenever you use someone else's computer, and run licensed software, you're committing copyright infringement. The decision split hairs distinguishing between the current user and the licensee for purposes of legality of making transient copies in memory as part of running the program. Peak was a repair business. MAI built computers (as in assembled/integrated; I think t…

I agree that MAI v Peek is crazy.

I was only pointing out that the law is of the opinion that a copy is a copy is a copy, regardless of where it's made, or how long it exists for.

Other decisions come into play to save us, like Authors Guild v Google, where they said search engines could make copies, bringing Fair Use into the picture.

Personally, I think that creating the model is Fair Use, but anything produced by the model would need to be checked for a violation. I would treat it the same as if I went to Google Book Search, and copied the snippet it returned into my new book.

The license associated with the training data then becomes insanely important. Having the model reference back to the source data is even more important.

For example, training data with a CC BY license would be very different to CC BY-SA and CC BY-ND, and they all require the work produced by the model to have credit back to the original source to be publishable.

https://creativecommons.org/licenses/

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

#377

Post by johnwalkr » Thu, Jun 01, 2023, 4:35 PM UTC

Loocid wrote: post #246

fomine3 wrote: post #242

Earlier quoted context omitted.

In my experience with Japan local IC/NFC credit card, I've never signed when I use CC in recent 5 years but PIN is sometimes needed. Do you use magnetic stripe CC?

>Do you use magnetic stripe CC. All my cards have an IC, however they were Australian cards. Perhaps it's an additional step for foreign credit/debit cards?

Must be a quirk for those cards in Japan. I've had no problems using my Canadian cards in Japan, and don't recall ever having to sign. In terms of banking and payments, it has improved a lot over the years: ATMs are mostly 24/7, or at least you can always find one, banks don't shut down over holiday weeks anymore (some do occasionally but it's a one-time thing for big upgrades for interoperability), and you can use credit cards and tap-to-pay almost everywhere. My only complaint is the sheer number of digital payment options. You end up with using at least 1-2 on top of your credit card because it's tied to some other service.

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

#378

Post by adam_arthur » Thu, Jun 01, 2023, 4:43 PM UTC

leni536 wrote: post #373

adam_arthur wrote: post #182

Earlier quoted context omitted.

Humans can memorize their training data too... aka see something and then produce a copy (code, drawing, music etc). The principles underlying how LLMs and humans learn isn't really that different... just different levels of loss/fuzziness.

And when humans do that they may also infringe on copyright.

Yet it’s not illegal to look at the McDonald’s logo, is it?

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

#379

Post by nologic01 » Thu, Jun 01, 2023, 4:51 PM UTC

stuckinhell wrote: post #366

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…

There are no such thing as "AI bros", AI training costs millions of dollars. It's a very different field than cryptocurrency where mining could be done by individuals in the early days. You see corporate tactics being deployed in a wide variety of ways to secure a social, legal, and economic moats, since apparently they have limited technology only based moat. It's similar to how google files various amicus briefs ag…

> There are no such thing as "AI bros", AI training costs millions of dollars

It is early days, in contrast with minting digital gold using software here the costs will come down. In any case not every "AI" model and application needs to pass Turing-test levels of language sophistry.

While you are right that (large) corporate interests are (and will be) the main actors here, I would not underestimate "noise traders" and other useful fools. With AI mania reaching apoplectic levels there is a wide population of actors that want to somehow get into the game.

The availability of open source AI models is an enabler in this respect, which may explain also the "leakage". Incidentally this issue is something the open source community needs to internalize and have a very clear position about.

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

#380

Post by johnwalkr » Thu, Jun 01, 2023, 5:20 PM UTC

lmm wrote: post #232

johnwalkr wrote: post #211

Earlier quoted context omitted.

Do you have concrete examples? I recently moved from Japan to Europe after 10 years in Japan and while some things seemed old-fashioned in Japan, things also change overnight there. During Covid most companies changed to digital signing. In Japan I often needed a paper from the city, but it was an easy-to-obtain printout I could get immediately from city hall or even from 7-11. I tend to require the same papers in Eu…

> Example 1: get ID card at the airport when you emigrate to japan. In Europe it takes 2-3 months. Yeah now have fun going through the process of renewing a My Number card. Go to city hall (during the hours when it's open), fill in a form, wait 2-3 months to get a notification that your replacement card has been made, then you have to book an appointment during office hours at city hall to pick it up (likely taking a…

I agree the my number card is a pain, if they want people to use it they should make it easier to get. When I went to get a plastic my number card, the staff at city hall advised me not to get it because I had a driving license and I'd have to renew the my number card every time my residence status was extended. They recommended it to non-holders of driving licenses to use as ID.

But I'm standing by these examples. For my move to Europe, it was 4 months to get the equivalent to the COE, then another 4 months (average is supposed to be 2-3 months) to get the equivalent of a residence card. And the company that sponsors you has to either prove you're highly skilled or prove they tried to hire a European for a set period of time first. And if you're not from a visa-waiver country, you aren't even allowed to travel in Europe while you wait for the ID card.

For my driving license, I probably won't have to wait in line, but the total time to process it is at least four months.

For starting a sole proprietorship, similar story. A bit annoying paperwork to start one in Japan, but it only took a few hours for everything. Here, it's 3 months and counting, and it's possible that it comes with a condition that I need to rent space for doing consulting (not do it from home) before getting the OK.

Japan is bureaucratic but usually fast, at least for individuals.