AI Art

[armorfiend]

You're overgeneralizing.

At minimum, Adobe Firefly is fully (and pointedly) trained on "Adobe Stock images, openly licensed content, and public domain content".

Given that they're openly promising to reimburse legal fees, I don't think they're bluffing. It'd only take one disgruntled insider, after all.

Click to shrink...

That does sound interesting. I tried to find out more about it but couldn't find anything on their site with a five minute search. Do you have a link?

Though as regards insiders, I mean, Boeing's guy is probably going to need a new job soon

 

[Shadlith]

That does sound interesting. I tried to find out more about it but couldn't find anything on their site with a five minute search. Do you have a link?

Click to shrink...

Literally on the page of Adobe Firefly:

Article:

Give creators practical advantages. Trained on Adobe Stock images, openly licensed content, and public domain content, Firefly is designed to be safe for commercial use. To ensure that creators can benefit from generative AI, we've developed a compensation model for Adobe Stock contributors whose content is used in the dataset to retrain Firefly models.

Source: https://www.adobe.com/products/firefly.html

And here's the page mentioning that it falls under their IP indemnification (for reference, that means that if a third party sues you, Adobe will compensate you for losses - they're saying their stuff is legal and putting their money where their mouth is.).

Article:

Responsibly designed to be safe for commercial use. Create with confidence knowing that our Firefly generative AI models were trained on licensed content, such as Adobe Stock, and public domain content where copyright has expired. We won't train our foundational Firefly models with your Custom Model training data. Customers on qualifying plans are eligible for IP indemnification for generated imagery (terms apply).

Source: https://www.adobe.com/creativecloud/business/enterprise/firefly.html

Standard "this is a business" caveats, obviously.

 

[Baughn]

Of course, using it does have the downside that you're supporting a major company that would like nothing more than to have every single artist pay them for access. You don't get to download the model, or experiment locally.

It's a bit funny how the anti-AI art crowd ends up supporting that sort of thing. But that's been mentioned plenty of times.

 

[Thellton]

Of course, using it does have the downside that you're supporting a major company that would like nothing more than to have every single artist pay them for access. You don't get to download the model, or experiment locally.

It's a bit funny how the anti-AI art crowd ends up supporting that sort of thing. But that's been mentioned plenty of times.

Click to shrink...

concur, at least with SD1.5, SDXL, and similar; I'm at least a dirty cyberpunk who isn't beholden to a corporation for continuing access to the AI in question unlike those corpo bootlickers who pay for their adobe subscription.

this was written tongue in cheek, but it is kind of accurate though, isn't it? the anti-AI crowd or rather the anti-open-source AI crowd crow about how it the model and it's training infringes upon the artist and their rights seemingly only when convenient, complaining about their images being scrapped when they 1) put in no effort to secure it themselves generally by looking into where was an appropriate space to exhibit their art, 2) post to corporates sites with Terms of Service that might as well say "you made this, but we own the copy" (Facebook was a big example of this a decade ago when they changed their ToS), and 3) complain about data scrapers scraping their data when they posted on sites that put in no effort to prevent scraping.

maddening...

 

[Mizu]

1) put in no effort to secure it themselves generally by looking into where was an appropriate space to exhibit their art, 2) post to corporates sites with Terms of Service that might as well say "you made this, but we own the copy" (Facebook was a big example of this a decade ago when they changed their ToS), and 3) complain about data scrapers scraping their data when they posted on sites that put in no effort to prevent scraping

Click to shrink...

I'd imagine it comes down to media difference around sharing. In reality people can't easily clone and otherwise use bits of your artwork, as you'd have to recreate it yourself by painting, drawing or moulding X material. On computers and the internet, near (?) every file is easily duplicated, and if you post artwork to X website there's nothing stopping someone from making a copy.

Before AI this just meant someone took your art and used it themselves, which kinda sucks. Now it has a bigger impact vs a single person, as the artbots use it as training data.

 

[armorfiend]

"Putting no effort into securing it"- the fuck how could they have? Like for real you're basically just saying it's their fault for having dared exist online.

Anyway. Adobe's new position is definitely interesting but I suspect it's got a lot of "we dare you to try to prove it" put in there. Still, it's a huge step in the right direction for you know actually compensating people who worked on things. Which is important because that's how you get more things.

 

[Morganite]

I mean… that's how AI art is generated and why so very many people are upset by it… because it's building off a massive database of stolen art, medical data, stock images, etc.

That's literally why copyright doesn't apply to it.

Click to shrink...

Um... no?

Seriously, I'm calling Citation Needed here. Because I've read the USCO's decision on the subject, and I don't remember it having anything to do with this. They were very clearly focused on the nominal lack of human authorship. And if you're talking about some other country, which one?

Also, the term would be something like 'infringed', not 'stolen'. The distinction does matter. (It may not be infringement either. Courts are moving slowly along... but I'm just saying, building large chunks of their case on incorrect assertions* may have been a bad plan.)

*Specifically, if the judges tentative rulings end up being what is ultimately issued, I don't think they've got much left that doesn't rely on the idea of "the model contains compressed copies of every training image", which I don't know what to say about it other than "Well, they've certainly got chutzpah".

-Morgan.

 

[Thellton]

"Putting no effort into securing it"- the fuck how could they have? Like for real you're basically just saying it's their fault for having dared exist online.

Anyway. Adobe's new position is definitely interesting but I suspect it's got a lot of "we dare you to try to prove it" put in there. Still, it's a huge step in the right direction for you know actually compensating people who worked on things. Which is important because that's how you get more things.

Click to shrink...

We all put stuff up online, far more than we probably should; and we all generally treat it with a very laissez-faire attitude because for most of us it's just words like these ones. but for artists and their art it is actually different. that's their portfolio that they've put up, a part of their resume; and they've collectively put in very little effort to protect their work as a profession. It is very much an exasperating attitude they have, and yes; they couldn't have predicted AI becoming capable of generating images, but you'd think they'd have done more to protect their portfolio just out of general principle.

 

[E.I.G.]

We all put stuff up online, far more than we probably should; and we all generally treat it with a very laissez-faire attitude because for most of us it's just words like these ones. but for artists and their art it is actually different. that's their portfolio that they've put up, a part of their resume; and they've collectively put in very little effort to protect their work as a profession. It is very much an exasperating attitude they have, and yes; they couldn't have predicted AI becoming capable of generating images, but you'd think they'd have done more to protect their portfolio just out of general principle.

Click to shrink...

So, most of the online artists I've seen who actually use art as a method of income are commission artists.
As in, those galleries are indeed their portfolio, which is used to attract and represent them to possible commissioners by being openly available online.

Even in the case of Patreon galleries locking some of it behind a paywall there is often a free gallery to show things off before such a payment.

 

[steamrick]

they couldn't have predicted AI becoming capable of generating images

Click to shrink...

They could've, actually. Various scientists and companies have been trying their hand at it for decades at this point.

It's just that no one gave serious thought to what the psychedelic daydreams of GANs in the 2010s might one day mean for them.

And starting with (arguably) GPT-3 in 2020 and Dall-E in 2021, the sheer speed at which all types of generative 'AI' improved as resources were poured in has caught just about everyone (including myself) completely off guard.

But no, this did not come out of nowhere.

 

[Kitsune]

I mean… that's how AI art is generated and why so very many people are upset by it… because it's building off a massive database of stolen art, medical data, stock images, etc.

That's literally why copyright doesn't apply to it.

So yeah if you are generating AI art you're by definition using someone else's work without their permission. Just 'cause you can't tell which bits were stolen from where doesn't mean it wasn't.

I can't be sure- I haven't asked- but I suspect most artists would prefer you stole an image outright and left their signature on it so people who saw it could at least look them up afterwards.

Click to shrink...

I can draw a little. Do you think I did not look at other people's drawings to try to see how other people made their drawings? I think most artists draw on what other artists have done in the past. The AI is effectively doing the same thing. May be more blatant but Royo is suppose to be inspired by Joaquin Sorolla, as an example.

 

[Baughn]

I can draw a little. Do you think I did not look at other people's drawings to try to see how other people made their drawings? I think most artists draw on what other artists have done in the past. The AI is effectively doing the same thing. May be more blatant but Royo is suppose to be inspired by Joaquin Sorolla, as an example.

Click to shrink...

The way humans learn has little to do with how AIs learn. Human learning acts more like in-context learning for AIs, which is barely applicable to image generators.

That difference is backwards directionally from how artists like to think of it, however. Humans get a lot more out of any single training examples than foundational AI training does; if there a copyright issues, they are dramatically lopsided against human learning. Except of course for the fact that copyright only triggers if you produce identical outputs; the procedure that gets you there is irrelevant.

AI is a whole lot less likely to produce identical outputs than humans are.

 

[Thellton]

So, most of the online artists I've seen who actually use art as a method of income are commission artists.
As in, those galleries are indeed their portfolio, which is used to attract and represent them to possible commissioners by being openly available online.

Even in the case of Patreon galleries locking some of it behind a paywall there is often a free gallery to show things off before such a payment.

Click to shrink...

The Patreon galleries are basically what I'd call optimal; because honestly, the fee that artists on Patreon are charging is essentially no different from a cover charge for going into a club (saying that I've never actually stepped foot in an actual club...) which is basically as close to a real art gallery as one is going to get in a digital medium. as an aside, I didn't realise Patreon had been around since 2013, time really has flown...

Anyway, the point that I'm making essentially is that it's daft to claim that one has total control over what is done with their work when they upload it to somewhere, when in most cases (baring Patreon) they don't even have control over who sees their art, let alone does with it after 'right click, save'.

actually now that I think about it but does sufficientvelocity have robots.txt? scratch that, sufficientvelocity and spacebattles both make use of the robots.txt which is good. both the "GPTBot" and "ChatGPT-User" user agents are OpenAI associated web crawlers. though it seems to me that it won't do jack against a web crawler from Meta, Google, Microsoft or really any other major corporation that one cared to name... nor a web crawler from a data hoarder either... well that is unless of course they've all secretly signed an agreement to standardise on an identifier?

 

[Tasrill]

concur, at least with SD1.5, SDXL, and similar; I'm at least a dirty cyberpunk who isn't beholden to a corporation for continuing access to the AI in question unlike those corpo bootlickers who pay for their adobe subscription.

this was written tongue in cheek, but it is kind of accurate though, isn't it? the anti-AI crowd or rather the anti-open-source AI crowd crow about how it the model and it's training infringes upon the artist and their rights seemingly only when convenient, complaining about their images being scrapped when they 1) put in no effort to secure it themselves generally by looking into where was an appropriate space to exhibit their art, 2) post to corporates sites with Terms of Service that might as well say "you made this, but we own the copy" (Facebook was a big example of this a decade ago when they changed their ToS), and 3) complain about data scrapers scraping their data when they posted on sites that put in no effort to prevent scraping.

maddening...

Click to shrink...

Man if I wanted to write a parody of what anti open source people say pro open source people think like I would have trouble topping this post. Sorry but 99.9% of open source projects don't scrape stuff off the internet without permission. They don't think 'well they didn't drm this so it is mine'. They are positively anal about proper attribution.

 

[Baughn]

which is barely applicable to image generators

Click to shrink...

Well, that statement lasted all of half a day. Actually it was incorrect when I wrote it.

GPT-4o (released yesterday) is a large-world-model AI with the native modalities of text (obviously), speech and images. In other words, it doesn't* need DALL-E to create or recognise images; it makes pictures the same way it makes text. What this means—what they've demonstrated—is that it can...

- Go from a longwinded text description of a character, to a picture of that character. "Longwinded description" could also mean "Ten chapters of a book."

- Use other pictures as reference inputs.

- Style transfer. "Like this picture, but cartoon."

- Character reference inputs, of course. "This character, but sitting at a table talking with this other character."

- Generalised edits. "Make the vase blue."

- Perfect consistency, up to and including generating GIFs orbiting some imagined object.

Basically everything I thought GPT-4-Turbo might have, but which it didn't. This time it's demonstrated capabilities; you can see it in their demo.

All of this is in-context learning, and its context window is 128k tokens at a minimum; at somewhere between a few hundred to a few thousand tokens per image, that means you can't fit an entire manga in the context window... but you certainly can fit references for all the characters, locations, and any other elements that need to remain consistent, and with a little bit of programming it would be almost trivially easy to build a manga-generator on this.

I... would have liked for this to be something that could be run locally. But as it stands, it looks like I'll be getting much better illustrations for my stories, moving forwards.

Oh and it's also about 5 times cheaper to run than GPT-4.

*: Though the image-output modality isn't available to anyone yet; they say it'll be rolled out over the next 2-3 weeks. If you ask 4o for an image right now it'll still use DALL-E, which is high quality but does almost none of this.

 

[Zoroastrians]

Recently I've been getting kinda annoyed about how people with SDXL can just quickly and easily produce similar (if rougher) content to what I can make with SD1.5 and a few hours of work. Then I realized I have no right to feel that way since, you know, what about real artists and how long it takes them? The people with actual talent.

Not relevant to the current, but I'd rather distract from the exact same conversation happening for a 12th time this year.

 

[Thellton]

Man if I wanted to write a parody of what anti open source people say pro open source people think like I would have trouble topping this post. Sorry but 99.9% of open source projects don't scrape stuff off the internet without permission. They don't think 'well they didn't drm this so it is mine'. They are positively anal about proper attribution.

Click to shrink...

99.9% of open source projects may not as you claim be scraping willy nilly, but it doesn't actually take every project scrapping, it just takes every project using the same map of useful scrapable resources (LAION) and scrapping them all in one go for their training run. something that artists if they had had control or influence over the sites that they posted to could have prevented or at least had leverage for saying "Hey I didn't consent to that". and yet both deviant art and art station are scrapable, unlike both Sufficient Velocity and Spacebattles for instance who at least put up some effort to stop dataset scraping. it seems to me that some of the significant places where artists have congregated to publish their art have most certainly not been acting in their user's interests unlike Patreon who do by my reading of their robot.txt a fairly comprehensive effort to prevent the big players from scraping their users' content.

So when I say it's 'ridiculous' for artists to shout and wail about their data being scrapped from where they posted it and being used to train image generation models, when many of them didn't put in effort to do basic safeguarding by considering who might even be seeing their work? and then they get upset after the fact? suffice to say, I am not impressed, as whilst they have good reason to be upset with the machine learning developers for using their work without consent and not considering how powerful things like using an artist's name in a Stable Diffusion prompt can be (a boring thing to do honest, if I want to see art by X artist; I'll search for art by X artist). I honestly don't recall hearing much of a peep from those same people wanting to tear apart deviant art and art station's owners for not using the robots.txt to safeguard their user data.

 

[E.I.G.]

99.9% of open source projects may not as you claim be scraping willy nilly, but it doesn't actually take every project scrapping, it just takes every project using the same map of useful scrapable resources (LAION) and scrapping them all in one go for their training run. something that artists if they had had control or influence over the sites that they posted to could have prevented or at least had leverage for saying "Hey I didn't consent to that". and yet both deviant art and art station are scrapable, unlike both Sufficient Velocity and Spacebattles for instance who at least put up some effort to stop dataset scraping. it seems to me that some of the significant places where artists have congregated to publish their art have most certainly not been acting in their user's interests unlike Patreon who do by my reading of their robot.txt a fairly comprehensive effort to prevent the big players from scraping their users' content.

So when I say it's 'ridiculous' for artists to shout and wail about their data being scrapped from where they posted it and being used to train image generation models, when many of them didn't put in effort to do basic safeguarding by considering who might even be seeing their work? and then they get upset after the fact? suffice to say, I am not impressed, as whilst they have good reason to be upset with the machine learning developers for using their work without consent and not considering how powerful things like using an artist's name in a Stable Diffusion prompt can be (a boring thing to do honest, if I want to see art by X artist; I'll search for art by X artist). I honestly don't recall hearing much of a peep from those same people wanting to tear apart deviant art and art station's owners for not using the robots.txt to safeguard their user data.

Click to shrink...

... I'm going to be blunt now, since you are being so clear on this:
You are victim blaming in about the same way as people who said artists with their work linked to NFTs should have minted the tokens themselves.
Fucking no it was not a reasonable precaution to prevent data scraping of this sort of thing before this all happened, because fucking nobody was misusing public image data in this way before then.

 

[Thellton]

... I'm going to be blunt now, since you are being so clear on this:
You are victim blaming in about the same way as people who said artists with their work linked to NFTs should have minted the tokens themselves.
Fucking no it was not a reasonable precaution to prevent data scraping of this sort of thing before this all happened, because fucking nobody was misusing public image data in this way before then.

Click to shrink...

well call me frustrated then, because no matter which way I look; I'm finding something to be dissatisfied with regards to the position of those who are arguing either position. Seriously, the only reason robots.txt are being used or not these days is because there is a valuation being made with regards to money on whether it's worthwhile or not, and the one who stands to profit or not is not the artist from this but rather the site owner. and unfortunately, that horse has already bolted from the paddock and is miles away at this point.

then there's the corporate image generators, Adobe may be fine because they've trained their model on their own stock image collection; and Stability.AI and other open source models may be fine too if only because 1) they're out in the wild and you can't refill pandoras box, and 2) because they're out in the open everyone can use them, including the artists who unwillingly contributed to their training. the only ones I can think of who stand on shaky ground at least from a moral consideration is the closed source model maintainers, particularly NovelAI and Midjourney; for obvious reasons.

basically, what I'm saying is, is that I've concluded that artists have 1) gotten hosed, 2) I actually don't see much chance of restitution, and 3) the things that could have been done to prevent them getting hosed were and are chillingly simple and yet the financial imperative of those who ran the sites was going to get in the way of that.

EDIT: So as it seems I'm coming off as victim blaming, that's my stupid fault for not being clear; because it really isn't my intention, as this is a problem of systems within systems, so essentially a microcosm of the problems of capitalism.

 

[E.I.G.]

well call me frustrated then, because no matter which way I look; I'm finding something to be dissatisfied with regards to the position of those who are arguing either position. Seriously, the only reason robots.txt are being used or not these days is because there is a valuation being made with regards to money on whether it's worthwhile or not, and the one who stands to profit or not is not the artist from this but rather the site owner. and unfortunately, that horse has already bolted from the paddock and is miles away at this point.

then there's the corporate image generators, Adobe may be fine because they've trained their model on their own stock image collection; and Stability.AI and other open source models may be fine too if only because 1) they're out in the wild and you can't refill pandoras box, and 2) because they're out in the open everyone can use them, including the artists who unwillingly contributed to their training. the only ones I can think of who stand on shaky ground at least from a moral consideration is the closed source model maintainers, particularly NovelAI and Midjourney; for obvious reasons.

basically, what I'm saying is, is that I've concluded that artists have 1) gotten hosed, 2) I actually don't see much chance of restitution, and 3) the things that could have been done to prevent them getting hosed were and are chillingly simple and yet the financial imperative of those who ran the sites was going to get in the way of that.

EDIT: So as it seems I'm coming off as victim blaming, that's my stupid fault for not being clear; because it really isn't my intention, as this is a problem of systems within systems, so essentially a microcosm of the problems of capitalism.

Click to shrink...

... so, there is a single problem with one part of what you are saying, and it is the core part of it all:
"the things that could have been done to prevent them getting hosed were and are chillingly simple"

Maybe to you these things seem simple, but I can very much assure you, and want a make a point to say to you, that most artists still do not know what robots.txt is or how it relates to the issue.
To the average artist, it is not only not clear what that means, it also isn't clear how they check for that or ask that it is in place. Which means without a clear reason to implement it they have no reason to have attempted to do so.

The issue is that you are arguing very bluntly and callously that the fix was "chillingly simple", and the truth is that it was not in fact that simple. It is in fact an obscure technical detail that they still might not know about. And your statements and arguments, even in this latest post, make it sounds like you think they should have had the prophetic vision to know that ahead of time.

My first reply to you was for a simple explanation to clarify why having a publicly viewable gallery was in fact a calculated risk. They have some or all of their artwork visible to everyone in order to promote those who like it to pay them to make more.
The proper precautions before this whole mess were things like signatures in the body of the artwork or watermarks. Things that make it harder for others to repost the artwork and claim it as their own. That was the reasonable precautions of the time, because that was the main risk: Other people reposting their work and claiming it.
As you are saying now, pandaora's box is open, and it wasn't until after that opening that we knew the cause of the issue. The concern was not there before then because the threat was not.

You are not wrong about anything else, but please stop thinking just because after the fact you know the technical side of the solution means that artists should have known about it beforehand.

 

[Thellton]

... so, there is a single problem with one part of what you are saying, and it is the core part of it all:
"the things that could have been done to prevent them getting hosed were and are chillingly simple"

Maybe to you these things seem simple, but I can very much assure you, and want a make a point to say to you, that most artists still do not know what robots.txt is or how it relates to the issue.
To the average artist, it is not only not clear what that means, it also isn't clear how they check for that or ask that it is in place. Which means without a clear reason to implement it they have no reason to have attempted to do so.

The issue is that you are arguing very bluntly and callously that the fix was "chillingly simple", and the truth is that it was not in fact that simple. It is in fact an obscure technical detail that they still might not know about. And your statements and arguments, even in this latest post, make it sounds like you think they should have had the prophetic vision to know that ahead of time.

My first reply to you was for a simple explanation to clarify why having a publicly viewable gallery was in fact a calculated risk. They have some or all of their artwork visible to everyone in order to promote those who like it to pay them to make more.
The proper precautions before this whole mess were things like signatures in the body of the artwork or watermarks. Things that make it harder for others to repost the artwork and claim it as their own. That was the reasonable precautions of the time, because that was the main risk: Other people reposting their work and claiming it.
As you are saying now, pandaora's box is open, and it wasn't until after that opening that we knew the cause of the issue. The concern was not there before then because the threat was not.

You are not wrong about anything else, but please stop thinking just because after the fact you know the technical side of the solution means that artists should have known about it beforehand.

Click to shrink...

Thinking of all this is very much after the fact even for me, but that's because I'm just one person; this whole Gen AI stuff literally came in like a whirlwind for me, and it's done the same for you and the rest of the planet. the problem is though is that artists as a professional group, who have unions for various subgroups across the planet (animators for example) should have been continuously watching for potential threats to their profession, and then collectively discussing how to adapt or engage with it. That doesn't seem to have happened at all though as they've been collectively blindsided by it entirely. So on that count, I feel somewhat surprised and disappointed that collective action dropped the ball so completely.

As to bluntly and callously, those who hold capital are blunt and callous, and no amount of nice words will ever be able to express concisely the simple, unfortunate truth. that in theory it was preventable, and that future artists work can at least be prevented from ending up in the datasets without their consent, if sites like deviant art and art station would do the same as Patreon with their Robots.txt which I've linked in a previous post.

as to "Maybe to you these things seem simple, but I can very much assure you, and want a make a point to say to you, that most artists still do not know what robots.txt is or how it relates to the issue.
To the average artist, it is not only not clear what that means, it also isn't clear how they check for that or ask that it is in place. Which means without a clear reason to implement it they have no reason to have attempted to do so." this is why talking and sharing knowledge is so important, why having collaborators is so important. the idea of an artist commune, which is sometimes the butt of jokes, or more seriously an artist's collective is and always will be a bloody good idea for precisely the above reason. none of us are omniscient, nor can a collective be so (case in point, digital artist's current plight), but it'd sure would have helped.

But as always, we come back to pandora... so as much as I might hand wring in sympathy for the problem they face and facepalm at their argumentation... there is very little to be done to take back what has happened. So I guess in the meantime, I'll continue using LLMs to help me code things that I otherwise couldn't, use a locally running Stable diffusion instance to create images that I couldn't afford to commission and firm up in my mind the image of characters in the stories I write that I don't have the guts to post Bah...

 

[Mizu]

Artists aren't gonna be tech savvy that often, not in the area of blocking their art getting copied by anyone who sees where they uploaded it. Most folk don't know what a robots text even is! I'm only vaguely aware from my programming course, and that's someone who actually works in website design.

As for getting blindsided, yeah, near every creative field seems to have been. Nobody can predict the future and five, seven years back, ai image or sound generation was laughable in quality. In the bad sense.

 

[steamrick]

The issue is that you are arguing very bluntly and callously that the fix was "chillingly simple", and the truth is that it was not in fact that simple. It is in fact an obscure technical detail that they still might not know about. And your statements and arguments, even in this latest post, make it sounds like you think they should have had the prophetic vision to know that ahead of time.

Click to shrink...

Saying that artists shouldn't be expected to know about robots.txt and other technical anti-scraper solutions and as such should be absolved of not using them is also problematic.

Taken further, this is the same train of thought as "they cannot be blamed for breaking the law because they didn't know such a law exsists".

The argument might not be as effective in a court of law as you seem to think it is.

Especially when combined with uploading their art to websites where doing so transfers rights and/or ownership. (And I'm explicitly not talking about third-party re-uploads.)
But I'm pretty sure we've had the Terms of Service debate before.

 

[Rgal]

Saying that artists shouldn't be expected to know about robots.txt and other technical anti-scraper solutions and as such should be absolved of not using them is also problematic.

Taken further, this is the same train of thought as "they cannot be blamed for breaking the law because they didn't know such a law exsists".

The argument might not be as effective in a court of law as you seem to think it is.

Especially when combined with uploading their art to websites where doing so transfers rights and/or ownership. (And I'm explicitly not talking about third-party re-uploads.)
But I'm pretty sure we've had the Terms of Service debate before.

Click to shrink...

Are you seriuosly comparing failure to protect one's interests with breaking a law? Bruh. That's taking victim blaming on a whole new level.

 

[steamrick]

Are you seriuosly comparing failure to protect one's interests with breaking a law?

Click to shrink...

I'll ask you kindly to not attempt to put words in my mouth.