Adobe caught selling AI generated images imitating Ansel Adams' work.

[ddutchison2]

Adobe agreed to take them down, eventually.

Adobe scolded for selling ‘Ansel Adams-style’ images generated by AI

 

[Mos6502]

I honestly feel like somebody being caught using AI generated images should be career ending. I was thinking back to the 2006 scandal where a photojournalist photoshopped some extra smoke into a picture of a warzone, and the image was quickly and rightly condemned, pretty much universally by photographers, reporters, and watchdogs. I wish some of that outrage and sense would return to us, instead of us acting like AI generated forgeries are just some bullsh!t we all need to tolerate so as not to hurt the feelings of corporations like Adobe. We're going to be overrun by AI generated fakes and frauds, and the companies pushing the tech really hope we won't make a fuss about it.

 

[olakiril]

It's important to note that it was a user not Adobe that had generated and sold the images on the platform.

While Adobe permits AI-generated images to be hosted and sold on its stock image platform, users are required to hold the appropriate rights or ownership over the content they upload. Adobe Stock’s Contributor Terms specifically prohibits content “created using prompts containing other artist names, or created using prompts otherwise intended to copy another artist.”

Adobe should be more careful though with the moderation.

Adobe Stock Vice President Matthew Smith previously told The Verge that the company generally moderates all “crowdsourced” Adobe Stock assets before they are made available to customers, employing a “variety” of methods that include “an experienced team of moderators who review submissions.”

Probably Adobe replaced the "experienced team of moderators" with AI 😀.

 

[Freakscene]

' Bassil Elkadi, Adobe’s Director of Communications and Public Relations, told The Verge that Adobe is “actively in touch with Ansel Adams on this matter” '

I guess they mean his estate unless they hired Steve Huff Huff Paranormal Who next?

 

[JohnGellings]

Adobe didn’t get “caught” trying to pass off AI as photography. They were informed someone was using Ansel Adam’s name to sell AI inspired images on its platform.

 

[Retro-Grouch]

This whole business is disgusting on so many levels...

 

[ddutchison2]

Adobe didn’t get “caught” trying to pass off AI as photography. They were informed someone was using Ansel Adam’s name to sell AI inspired images on its platform.

Yes, but Adobe were informed multiple times by the legitimate owners of those images - over a period of a year, by their account - before they agreed to follow their own rules and take them down - hence my choice of the word "caught". Had they reacted immediately and correctly, there'd be no story.

 

[tcmx3]

I honestly feel like somebody being caught using AI generated images should be career ending. I was thinking back to the 2006 scandal where a photojournalist photoshopped some extra smoke into a picture of a warzone, and the image was quickly and rightly condemned, pretty much universally by photographers, reporters, and watchdogs. I wish some of that outrage and sense would return to us, instead of us acting like AI generated forgeries are just some bullsh!t we all need to tolerate so as not to hurt the feelings of corporations like Adobe. We're going to be overrun by AI generated fakes and frauds, and the companies pushing the tech really hope we won't make a fuss about it.

tbh if someone wants to use "AI" (a laughable term for these models) that's like, theoretically OK to me for stock photos.

however, and this is a load bearing 'however', these models are only possible with theft. it takes massive amounts of money (and energy) to train generative models and the companies with those resources have shown time and time again they need to be taken out back behind the shed and {redacted} because they just steal and steal and steal and it is only through inaction and regulatory capture that they can continue to operate. they are all, without exception, run by the worst person you know. so while I think the photojournalism analogy is somewhat stretched (though I strongly agree that folks in that field should not use it ever) since some folks are not doing reportage, in practice I agree.

use AI? you're done. gone. sleeping with the (proverbial) fishes.

make the person selling the images do a few hours of community service at a museum and the company that released the model pay RIAA vs Jammie Thomas-Rasset money IMO

 

[boojum]

I honestly feel like somebody being caught using AI generated images should be career ending. I was thinking back to the 2006 scandal where a photojournalist photoshopped some extra smoke into a picture of a warzone, and the image was quickly and rightly condemned, pretty much universally by photographers, reporters, and watchdogs. I wish some of that outrage and sense would return to us, instead of us acting like AI generated forgeries are just some bullsh!t we all need to tolerate so as not to hurt the feelings of corporations like Adobe. We're going to be overrun by AI generated fakes and frauds, and the companies pushing the tech really hope we won't make a fuss about it.

Money talks.

 

[boojum]

' Bassil Elkadi, Adobe’s Director of Communications and Public Relations, told The Verge that Adobe is “actively in touch with Ansel Adams on this matter” '

I guess they mean his estate unless they hired Steve Huff Huff Paranormal Who next?

"Actively in touch"? OK, what does that mean in English rather than doublespeak? These are typical corporate platitudes. They go hand in hand with other platitudes and people swallow them whole without question. I fear this is just the beginning rather than the ending.

 

[JohnGellings]

Yes, but Adobe were informed multiple times by the legitimate owners of those images - over a period of a year, by their account - before they agreed to follow their own rules and take them down - hence my choice of the word "caught". Had they reacted immediately and correctly, there'd be no story.

Who are the legitimate owners of the "images?" Seems to me it isn't about the images, it is about using Ansel Adams´name attached to the images.

 

[tcmx3]

Who are the legitimate owners of the "images?" Seems to me it isn't about the images, it is about using Ansel Adams´name attached to the images.

Article states Adobe won't let you upload images with prompts that have the artists name in it.

I personally, trust that statement about as far as I can throw a life sized gold statue of the adobe CEO.

Here's an example of a company that totally wasn't training their model specifically to spit out results based on artist names that only came out during discovery when they were sued: Leaked: the names of more than 16,000 non-consenting artists allegedly used to train Midjourney’s AI

 

[boojum]

This is just a small example.

 

[olakiril]

these models are only possible with theft.

I wouldn't go that far. Our brain also operates in a fairly similar manner and as humans we imitate. A lot. It would be quite easy for a photographer to create images in the style of Ansel Adams and plenty do so and with many different styles.
In this case the ethical issues are:
a) Using the artist's name for personal gain. Specifically, had the Adobe Stock user avoided Ansel Adams name it wouldn't have been a problem as these pictures don't look like actual pictures of AA.
b) Adobe allowing this. I guess Ansel Adams was never a subscriber and thus didn't "exist" as an artist for Adobe.

 

[JohnGellings]

Article states Adobe won't let you upload images with prompts that have the artists name in it.

Well, it must be hard to police right now... once informed, they removed them. Unfortunately, the sheer amount of people using these products that is going to happen.

I personally, trust that statement about as far as I can throw a life sized gold statue of the adobe CEO.

Yeah, well, it is an easy target.

Here's an example of a company that totally wasn't training their model specifically to spit out results based on artist names that only came out during discovery when they were sued: Leaked: the names of more than 16,000 non-consenting artists allegedly used to train Midjourney’s AI

Yeah, and? That is not what is happening here.

 

[SWB]

Plagiarism is one thing but it is possible to take umbrage too far and start seeing things that aren't there. Next we'll have to be careful saying who it is we're inspired by in our photography before people accuse us of copying. It was Picasso who said 'good artists copy, great artists steal!' meaning I guess that you take inspiration from somebody and make it your own.

 

[Mos6502]

I wouldn't go that far. Our brain also operates in a fairly similar manner and as humans we imitate. A lot. It would be quite easy for a photographer to create images in the style of Ansel Adams and plenty do so and with many different styles.

Our brain does not operate in a fairly similar manner. This is a line handed out by AI buffs with absolutely nothing to substantiate it. Sam Altman of OpenAI recently admitted that he doesn't really understand how AI works.

However, there's one obvious difference between how humans sense and create, and how computers do. As humans we take in "data" through our combined senses, every day, hour, minute, second, our whole life through, and all of this plays a part in how we interact with the world, and informs our conscious decisions and subconscious processes. With AI, the only data that is taken in is what is specifically given. Computers have no other inputs than what we give them. Even when creating visual art, a person draws on all sorts of non-visual inspiration, AI doesn't, and can't (yet, anyway).

As for the theft question... there's really no question about it. The LAION dataset contains over five billion images. Used, of course, without permission, in a wholly unprecedented manner, on a scale never before seen*. Maybe that constitutes theft, maybe it doesn't, but it does make it clear that AI takes a lot of work from others who get no credit or compensation. What it also make obvious is that AI needed literally decades worth of material to be brought up to the state of perfectly making bad propaganda to fool boomers on facebook. All the human creativity available on the internet, across the globe, and silicon valley has turned it into mindless sh!t. 🤔

*= further it has been shown that some materials in this dataset have been obtained illegally, or are images which contain illegal content.

 

[tcmx3]

I wouldn't go that far. Our brain also operates in a fairly similar manner and as humans we imitate. A lot. It would be quite easy for a photographer to create images in the style of Ansel Adams and plenty do so and with many different styles.
In this case the ethical issues are:
a) Using the artist's name for personal gain. Specifically, had the Adobe Stock user avoided Ansel Adams name it wouldn't have been a problem as these pictures don't look like actual pictures of AA.
b) Adobe allowing this. I guess Ansel Adams was never a subscriber and thus didn't "exist" as an artist for Adobe.

oh for sure I love when I'm looking at an Ansel Adams photograph and my brain is all like convert the luminances of the red channel green channel and blue channel of this image into numerical values between 0 and 255 and then rescale them between 0 and 1, downsize them to a say 64x64x3 tensor, randomly generate an additional 64x64x3 tensor and then convolve over it to compare 3x3 matrices of my random inputs against said downsized photograph, measure the distance via binary cross entropy, back propagate the differences to update the weights and biases of my transformation matrix against which I multiply my randomly generated tensor and then repeat this 10,000 times per second on my vast hoard of images I am using without consent for a purpose never imagined by the original artist.

probably my favorite way to spend a Tuesday if I'm honest.

 

[tcmx3]

edit: nm a bit too off topic

 

[Mos6502]

I do remember one of the big wigs at Stability AI stated publicly a couple years ago (on twitter) that training data was running out, and that there was a lot of material behind firewalls that people should try and get their hands on... so no I don't really trust any of the people behind the technology. There is nobody in that space that seems capable of working ethically. Indeed, we've seen time and again that those who wish to do so end up getting forced out of it.