|
Owling Howl posted:Moreover others will make these systems available no matter what. It doesn't really help an artist that Microsofts system isn't trained on his art while a Russian or Chinese system is flooding the internet with infinity "in the style of" images. That the damage was done by shithead techbros already is no argument to double down, whether it was Uber or this. Now we can at least try and stem it somewhat.
|
#
?
Apr 5, 2023 00:27
|
|
|
|
| # ? May 26, 2024 10:45 |
|
|
Going the copyright route is less stemming the flow and more like trying to cover one of the outflows in a Y valve with a finger, you're just redirecting it to Adobe, Disney et al. How nice for them, I'm sure they'll be responsible with their monopoly.
|
|
#
?
Apr 5, 2023 00:35
|
|
|
SCheeseman posted:Going the copyright route is less stemming the flow and more like trying to cover one of the outflows in a Y valve with a finger, you're just redirecting it to Adobe, Disney et al. How nice for them, I'm sure they'll be responsible with their monopoly. As said before, uncopyrightable content is useless to those folks. There will always be dipshit Russian internet lawbreakers, that's not an argument against laws.
|
|
#
?
Apr 5, 2023 00:40
|
|
|
StratGoatCom posted:AI content CANNOT be copyrighted, do you not understand? Nonhuman processes cannot create copyrightable images under the current framework. You can copyright human amended AI output, but as the base stuff is for all intents and purposes open domain, if someone gets rid of the human stuff, it's free. The inputs do not change this. Useless for anyone with content to defend. And it would be a singularly stupid idea to change that, because it would allow a DDOS in essence against the copyright system and bring it to a halt under a wave of copyright trolling. One thing I don't really understand here, is what the legal difference is between text interpreted computer generated works and mouse click interpreted generated works? My zero legal education impression is that I could open Paint, tell it I want a 400x600 picture with a star in each corner and the text, "Barry's Mowing Service" written in Comic sans in the middle and I could secure a copyright on that image. If I was to ask a home-ran image generator to do the same thing, using incredibly precise language rather than mouse clicks to tell it what to draw and what font to use, would that be okay under current law? Is it based on the precision of the human input? The randomness of the AI input? Or something else?
|
|
#
?
Apr 5, 2023 00:46
|
|
|
Gentleman Baller posted:One thing I don't really understand here, is what the legal difference is between text interpreted computer generated works and mouse click interpreted generated works? A human was controlling the mouse.
|
|
#
?
Apr 5, 2023 00:54
|
|
|
StratGoatCom posted:As said before, uncopyrightable content is useless to those folks. There will always be dipshit Russian internet lawbreakers, that's not an argument against laws. If you're talking about that recent case where AI generated works were ruled uncopyrightable, that doesn't mean that you aren't able to use what was generated in a greater work and not have that be copyrightable. Same rules as public domain works. If they were to inherit the copyright of everything in the dataset, this still makes fully licensed datasets legal, which has the knock-on effect of entrenching current corporate IP monopolies while not doing much to prevent layoffs from automation. If you want works to be uncopyrightable and any works based on them uncopyrightable, I don't think there is any precedent for this. It's also dangerous, given that identifying if a work is derived from AI generated output would be difficult and nebulous, particularly if it was modified and copyright being copyright "it was AI generated!" will become another legal tool wielded by IP holders. Forced watermarking? Well, watermarks are probably also going to be on the chopping block soon; they're designed to be hard for human eyes to see, AI can almost certainly be trained to identify, recognize and remove them.
|
|
#
?
Apr 5, 2023 00:55
|
|
|
SCheeseman posted:
The issue is that you can at least potentially pull that poo poo out of the human amended work, and be untouchable. It's a radioactive mess, vis a vis chain of title, and btw? It's not unprecedented, it is literally how it is for nonhuman done work already.
|
|
#
?
Apr 5, 2023 00:58
|
|
|
StratGoatCom posted:A human was controlling the mouse. Yeah, but my question is, what is the current legal difference between that and a human controlling the keyboard, if the human controlling the keyboard provided the exact hex code of every colour, text size, pixel location if every star etc. Would that be the same under current law or what?
|
|
#
?
Apr 5, 2023 01:00
|
|
|
Gentleman Baller posted:Yeah, but my question is, what is the current legal difference between that and a human controlling the keyboard, if the human controlling the keyboard provided the exact hex code of every colour, text size, pixel location if every star etc. Would that be the same under current law or what? No, because a human did it.
|
|
#
?
Apr 5, 2023 01:07
|
|
|
StratGoatCom posted:The issue is that you can at least potentially pull that poo poo out of the human amended work, and be untouchable. It's a radioactive mess, vis a vis chain of title, and btw? It's not unprecedented, it is literally how it is for nonhuman done work already. Sure, if everyone distributed everything in PSD format with unmerged layers. Even a relatively small change is enough to turn it into a derivative work. You're way overblowing the consequences of using uncopyrightable works in a greater work, something Disney has made billions of dollars doing. A work created by a human that includes non-human works doesn't have it's copyright invalidated by the non-human work.
|
|
#
?
Apr 5, 2023 01:08
|
|
|
Gentleman Baller posted:Yeah, but my question is, what is the current legal difference between that and a human controlling the keyboard, if the human controlling the keyboard provided the exact hex code of every colour, text size, pixel location if every star etc. Would that be the same under current law or what? I wrote a brief piece on this in the GBS AI Art thread. KwegiboHB posted:Well the short answer is: you're good. Both projects are safe. This thread has a lot of good conversation going on in it. I have quite a few things I can give insight on but I refuse to rush through any of this. There is a lot to untangle and some of these topics require serious consideration of multiple angles and that just takes time. Time I'm going to take before I post in here. Looking forward to it though. https://hub.jhu.edu/2023/02/28/organoid-intelligence-biocomputers/ AHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
|
|
#
?
Apr 5, 2023 01:10
|
|
|
KwegiboHB posted:I wrote a brief piece on this in the GBS AI Art thread. Nice, exactly what I was wondering about, thanks. Very interesting.
|
|
#
?
Apr 5, 2023 01:15
|
|
|
SCheeseman posted:Sure, if everyone distributed everything in PSD format with unmerged layers. Even a relatively small change is enough to turn it into a derivative work. You're way overblowing the consequences of using uncopyrightable works in a greater work, something Disney has made billions of dollars doing. By doing a lot of visual asset work and their own writing. if you're gonna process it fairly heavily to the point where you're sure it's nothing extractable of the original work for say, an AI, at that point what was even the point of using the AI in the first place? StratGoatCom fucked around with this message at 01:26 on Apr 5, 2023 |
|
#
?
Apr 5, 2023 01:23
|
|
|
SCheeseman posted:Impede, obstruct, whatever. In any case what you want isn't going to make AI art generators uneconomical, it'll make the 'legal' ones economical only for the entrenched IP hoarders. What you want changes nothing about how people will in actuality be exploited and may even serve to make it worse! I don't know what the heck you're responding to, but it's not anything I said. At no point did I propose removing exploitation from the entire media industry as a whole. I'm not pushing some grand utopian reworking of media as we know it, nor am I talking about ways to remove all exploitation from media production. I honestly have no idea what you're talking about. All I said was that AI companies should have to follow current law and respect the existing rights of media creators and media owners. Yes, that won't put a total end to all exploitation of creatives. Nowhere did I say that it would! Honestly, that doesn't even make much sense as a response. Personally, I don't give a poo poo if Disney trains an AI dedicated to spitting out pictures of Mickey Mouse, as long as they only use art they have the right to use. I don't care if they feed all nine Star Wars movies into an AI created for the sole purpose of making a new Star Wars movie every two weeks. It would be a very foolish thing for them to do, but if they're really that determined to shoot themselves in the feet, they have every right to do so. As long as they're only using content they own, and aren't scraping fansites for Luke x Chewbacca slash fanfiction to beef up their dataset, I don't see much real harm from it, except to the very brands they use this on. Gentleman Baller posted:One thing I don't really understand here, is what the legal difference is between text interpreted computer generated works and mouse click interpreted generated works? The randomness of the AI output. As the Copyright Office put it, "prompts function closer to suggestions than orders, similar to the situation of a client who hires an artist to create an image with general directions as to its contents".
|
|
#
?
Apr 5, 2023 01:25
|
|
|
StratGoatCom posted:By doing a lot of visual asset work and their own writing. if you're gonna process it fairly heavily to the point where you're sure it's nothing extractable of the original work for say, an AI, at that point what was even the point of using the AI in the first place. I'm physically disabled.
|
|
#
?
Apr 5, 2023 01:26
|
|
|
KwegiboHB posted:I'm physically disabled. I have diagnosed motor issues among other problems and draw a disability pension, I don't touch that poo poo because I am not a person who shits on the disabled who make a living there.
|
|
#
?
Apr 5, 2023 01:27
|
|
|
StratGoatCom posted:I have diagnosed motor issues among other problems and draw a disability pension, I don't touch that poo poo because I am not a person who shits on the disabled who make a living there. Your angry, I get that. You asked a question and I answered.
|
|
#
?
Apr 5, 2023 01:30
|
|
|
StratGoatCom posted:By doing a lot of visual asset work and their own writing. if you're gonna process it fairly heavily to the point where you're sure it's nothing extractable of the original work for say, an AI, at that point what was even the point of using the AI in the first place. The main use case for this in business and office is as a production accelerant, replacement for stock artwork and for img2img-like applications. The stuff that the generator spits out raw is rarely useful in of itself, more so as part of a production pipeline. SCheeseman fucked around with this message at 01:36 on Apr 5, 2023 |
|
#
?
Apr 5, 2023 01:33
|
|
|
Main Paineframe posted:I don't know what the heck you're responding to, but it's not anything I said. At no point did I propose removing exploitation from the entire media industry as a whole. I'm not pushing some grand utopian reworking of media as we know it, nor am I talking about ways to remove all exploitation from media production.
|
|
#
?
Apr 5, 2023 01:34
|
|
|
SCheeseman posted:I know. You should, though. Then you want these scraped models nuked flat. If they can't pay or otherwise secure permission, then they can go to hell if 'fairness' is to mean anything beyond computer touchers getting cheap toys.
|
|
#
?
Apr 5, 2023 01:37
|
|
|
KwegiboHB posted:https://hub.jhu.edu/2023/02/28/organoid-intelligence-biocomputers/
|
|
#
?
Apr 5, 2023 01:42
|
|
|
StratGoatCom posted:Then you want these scraped models nuked flat. If they can't pay or otherwise secure permission, then they can go to hell if 'fairness' is to mean anything beyond computer touchers getting cheap toys.
|
|
#
?
Apr 5, 2023 01:43
|
|
|
SCheeseman posted:That mass societal damage happens over and over again as a result of automation indicates to me that the problem sits higher up the chain and probably isn't solvable by using a framework created by a cartel of IP holders to further entrench their monopolies. This isn't some ip cartel, this is enforcement of long standing rights in law in basically every legal system. The scraped material was under copyright from the instant the job was done.
|
|
#
?
Apr 5, 2023 01:45
|
|
|
StratGoatCom posted:A human was controlling the mouse. And a human is running the AI But you're somewhat right that, currently, purely AI generated stuff doesn't seem to be copyrightable because it doesn't have enough human creative input. Though if it can have enough, and where that line is, is currently unknown (but being tested right now and I suspect we will find out sooner than later now that adobes implemented AI stuff into Photoshop) Like, if I draw an image and dump it into img2img, is that copyrightable? We don't know. My Deforum videos that rely heavily on math to direct the AI into animating things? We have no idea. Etc. There is a controlnet picture up for application right now I think, so we'll see how that goes And it's not because an AI can't copyright anything. That was the justification in the Thaler case because he refused at every step of the way to acknowledge a human was at all involved. He wasnt trying to actually get a copyright but to make a statement and push copyright law to its limits like he had done with some other things. And in that ruling they explicitly said "we could be testing what level of human involvement is necessary but we're not because he wants the AI to be listed as the owner of the work and, so, non-human things just can't do that." In all other, real AI copyright scenarios the person running the AI is trying to get copyright, not the AI, and so the nonhuman nature of AI doesn't matter.
|
|
#
?
Apr 5, 2023 01:46
|
|
|
StratGoatCom posted:This isn't some ip cartel, this is enforcement of long standing rights in law in basically every legal system.
|
|
#
?
Apr 5, 2023 01:47
|
|
|
cat botherer posted:the answer is not to increase the power of IP holders. For the hundedth time, this isn't. This is merely using already existing rules. Allowing it to be otherwise in fact will have that effect you fear, because it makes literally anything free real estate for billionaire bandits. Indeed, the point is laundering this behavior, much as crypto was laundering for securities bs. quote:And a human is running the AI So? It's no different then throwing dice, and we don't allow copyright of stuff from that either. Procedures are not copyrightable.
|
|
#
?
Apr 5, 2023 01:51
|
|
|
StratGoatCom posted:This isn't some ip cartel, this is enforcement of long standing rights in law in basically every legal system. There is no natural right to IP. The earliest cases of IP enforcement hundreds of years ago mostly served the aristocracy and the past century has seen massive changes in how and what rights are granted, when they're taken away and dealing with the many edge cases. There is absolutely a cartel, with all the familiar names lobbying US government to amend copyright law to be more favorable to their business with the US using soft power via trade agreements to push the cartel's vision of copyright across the globe.
|
|
#
?
Apr 5, 2023 01:55
|
|
|
StratGoatCom posted:So? It's no different then throwing dice, and we don't allow copyright of stuff from that either. Procedures are not copyrightable. Well, like I said, that's why we can't just say AI generated stuff is uncopyrightable on the grounds that an AI is not a human. That's not the argument of the USCO for the vast majority of copyright registration attempts and that's the reason there is probably some line of human interaction where it is copyrightable. Like the example of the current thing up for copyright, using controlnet is not throwing dice but will it be non-random enough? The only thing we can actually say right now is "we don't know"
|
|
#
?
Apr 5, 2023 01:55
|
|
|
Not 'probably' there's some line where it's copyrightable, we already know there is in fact a line where it's copywritable because it's already established to be somewhere below 'arrangement into the panels of a comic book,' it's just the exact place of the line that's in dispute.
|
|
#
?
Apr 5, 2023 01:58
|
|
|
StratGoatCom posted:For the hundedth time, this isn't. This is merely using already existing rules. Allowing it to be otherwise in fact will have that effect you fear, because it makes literally anything free real estate for billionaire bandits. Indeed, the point is laundering this behavior, much as crypto was laundering for securities bs. You sure about that? I think that if someone uses a noise tool to generate a 3D landscape, they can copyright that, and that's fancy dice. Or, am I wrong about that?
|
|
#
?
Apr 5, 2023 01:59
|
|
|
StratGoatCom posted:For the hundedth time, this isn't. This is merely using already existing rules. Allowing it to be otherwise in fact will have that effect you fear, because it makes literally anything free real estate for billionaire bandits. Indeed, the point is laundering this behavior, much as crypto was laundering for securities bs.
|
|
#
?
Apr 5, 2023 01:59
|
|
|
reignonyourparade posted:Not 'probably' there's some line where it's copyrightable, we already know there is in fact a line where it's copywritable because it's already established to be somewhere below 'arrangement into the panels of a comic book,' it's just the exact place of the line that's in dispute. Well, in that case the actual images in the panels are not copyrighted. Which I think a lot of people wouldn't consider a big enough protection, at least for commercial use. I can still take the Zarya of the Dawn images and do what I want with them because they are the part of that that didn't get copyright protection
|
|
#
?
Apr 5, 2023 02:00
|
|
|
SCheeseman posted:There is no natural right to IP.  poo poo you'd know if you knew even the slightest bit of copyright law, in america, canada or the EU in order posted:Copyright protection automatically exists as soon as a work is created and fixed in a tangible medium of expression. Protection applies to works at any stage of creation (from initial drafts to completed works) and to published and unpublished works... 
|
|
#
?
Apr 5, 2023 02:02
|
|
|
Those are legal rights.
|
|
#
?
Apr 5, 2023 02:03
|
|
|
SCheeseman posted:Those are legal rights. So? By training those models, they clearly crossed long established lines on copyright law.
|
|
#
?
Apr 5, 2023 02:05
|
|
|
You said IP had long standing rights in law in basically every legal system, which I took for you to mean that it's a natural law with your talk of long standing, universal rights. But in any case they aren't long standing and certainly haven't been universal. Night of the Living Dead is famously public domain in spite of the assumption that copyright is always automatic and, more relevant, we even impose limits based on the minimum amount of human effort and intent required for something to qualify.
|
|
#
?
Apr 5, 2023 02:14
|
|
|
SCheeseman posted:You said IP had long standing rights in law in basically every legal system, which I took for you to mean that it's a natural law with your talk of long standing, universal rights. So? Those rights are automatic for actual work, instead of machine output unless otherwise. They were violated quite openly on the assumption that artists could not exercise them.
|
|
#
?
Apr 5, 2023 02:17
|
|
|
StratGoatCom posted:So? By training those models, they clearly crossed long established lines on copyright law.
|
|
#
?
Apr 5, 2023 02:26
|
|
|
StratGoatCom posted:So? Those rights are automatic for actual work, instead of machine output unless otherwise. Except in cases where they aren't, given the jurisdiction and timeframe we're talking about. Right now the jurisdiction is the US in a time government is (mostly) favorable to big tech. Given court outcomes that find a fair use argument compelling, that precedent is likely to spread beyond the US' borders. You'll probably find it harder to defer to copyright law as if it's an arbiter of ethics once that has happened.
|
|
#
?
Apr 5, 2023 02:26
|
|
|
|
| # ? May 26, 2024 10:45 |
|
|
cat botherer posted:They haven't though. Training models on copyrighted things is nothing new. It's been going on for well over a decade. If it was crossing established lines, there would be case law on it by now. Can you actually point to evidence of your legal theories? Given the commercial nature of these models and that they create similar outputs WITHOUT permission, no I do not think fair use harbor applies. SCheeseman posted:Except in cases where they aren't, given the jurisdiction and timeframe we're talking about. Right now the jurisdiction is the US in a time government is (mostly) favorable to big tech. Given court outcomes that find a fair use argument compelling, that precedent is likely to spread beyond the US' borders. You'll probably find it harder to defer to copyright law as if it's an arbiter of ethics once that has happened. That is debatable, given the long standing treaties this would violate, not to mention things like EU law which may well be less charitable.
|
|
#
?
Apr 5, 2023 02:28
|
|