I pay OpenAI for a chat and image generation service. If I make Mario or something random, I pay them the same amount. If I go sell those pictures of Mario that I made with the service then I am liable for infringement, not OpenAI. OpenAI is not charging me more for making Mario or anything else.
Same as if I draw Mario to keep privately or draw him and sell the images. Adobe is never mentioned as a liability even though I used that software to infringe and paid Adobe for the ability to do so.
Please tell me how it’s different. Don’t tell me scale because they don’t care if it’s one or 1 million Marios. If someone was making money on a million Marios they would be sued independently, whether or not they used AI.
The short version is that it’s a licensing issue. All art is free to view, but the moment you try to integrate it into a commercial product/service you’ll owe someone money unless the artist is given fair compensation in some other form.
For example, artists agree to provide a usage license to popular art sites to host and display their works. That license does not transfer to the guy/company scraping portfolios to fuel their AI. Unfortunately, as we can see from the article, AI may be able to generate but it still lacks imagination and inspiration; traits fundamental to creating truly derivative works. When money exchanges hands that denies the artist compensation because the work was never licensed and they are excluded from their portion of the sale.
Another example: I am a photographer uploading my images to a stock image site. As part of ToS I agree to provide a license to host, display, and relicense to buyers on my behalf. The stock site now offers an AI that create new images based on its portfolio. The catch is that all attributed works result in a monetary payment to the artists. When buyers license AI generated works based on my images I get a percentage of the sale. The stock site is legally compliant because it has a license to use my work, and I receive fair compensation when the images are used. The cycle is complete.
It gets trickier in practice, but licensing and compensation is the crux of the matter.
At some point these companies will need to get licenses for any copyrighted work that was part of the training data, or start over with public domain works only. The art may be data, but that data has legal owners whose rights grant control over it’s use.
Another way to think about is proprietary code. You can see it and learn from it at your leisure. But to use it commercially requires a license, one that clearly defines what can and cannot be done with it, as well as fair compensation.
But it’s not reposting copyrighted images. It is analyzing them, possibly a long time ago, then using complex math and statistics to learn how to make new images when requested, on the fly. It’s an automated version of the way humans learn how to make art or take pictures. If it happens to produce Mario very closely it is because it learned very well.
That is why this isn’t cut and dry. And why it might be good to think of it as derivative works. I don’t think you will be able to nail down this idea of imagination and inspiration. It’s just not that straight forward.
Edit: Also, the generator is not pumping out copyrighted images intentionally. It is waiting for a prompt from a user. Who will then go and post it somewhere. If it is too close to Mario, it is that human user who has violated copyrights. They only used the generator as a tool. I feel like that is very relevant.
That’s the ideal and how it’s advertised, but in reality they retain a lot more of the original copies than most people think, in addition to the fact that the output is often not sufficiently “transformative” in copyright terms to avoid being considered a derivative work still needing a copyright license.
In your Mario example, the character as such is unique enough that it has its own copyright and you can’t use images of that character commercially without a license regardless of how the image was created. If it’s recognizable as Mario then you copied the design as far as a judge would be concerned. If you asked a human to draw it then it would be equally infringing.
A human doesn’t even need to ask for a copyrighted work for it to generate infringing outputs.
I don’t think intention or prompting matters much. If you type “Mario movie” into the YouTube search box and it shows you the Mario movie, YouTube needs to license that material, even if you explicitly ask it to do so and even if you don’t redistribute it. An AI tool is in a similar situation, you still need to license content if you’re making a tool.
I think the problem is that the user often doesn’t realize that they’re infringing on copyright because the prompts don’t specifically mention the copyrighted work.
In other word, there is a need for a way to tell the AI that only public domain work can be used in the generation.
Very possible people don’t realize. And you know what? We shouldn’t care. But if someone generates a Mario and puts in on their website or makes a fanfic comic, doesn’t matter how they made it… go after that person. Just like you always have, Nintendo…
But I worry for the future of any tool if they win this. Add a feature to a computer art tool that feels too “generatey”, you better watch out… I worry about human artists, having to prove the sources they learned from were not protected copyrights when they lean into a style that feels like Nintendo’s…
I’m not an infrengment lawyer… but disney and nintendo, and NYtime and whole lot of artist seems to think they have a case.
I suspect it is the same as using a sample from a beyonce song in something you are selling, you may have a problem with beyonce’s jurists
I pay OpenAI for a chat and image generation service. If I make Mario or something random, I pay them the same amount. If I go sell those pictures of Mario that I made with the service then I am liable for infringement, not OpenAI. OpenAI is not charging me more for making Mario or anything else.
Same as if I draw Mario to keep privately or draw him and sell the images. Adobe is never mentioned as a liability even though I used that software to infringe and paid Adobe for the ability to do so.
Please tell me how it’s different. Don’t tell me scale because they don’t care if it’s one or 1 million Marios. If someone was making money on a million Marios they would be sued independently, whether or not they used AI.
The short version is that it’s a licensing issue. All art is free to view, but the moment you try to integrate it into a commercial product/service you’ll owe someone money unless the artist is given fair compensation in some other form.
For example, artists agree to provide a usage license to popular art sites to host and display their works. That license does not transfer to the guy/company scraping portfolios to fuel their AI. Unfortunately, as we can see from the article, AI may be able to generate but it still lacks imagination and inspiration; traits fundamental to creating truly derivative works. When money exchanges hands that denies the artist compensation because the work was never licensed and they are excluded from their portion of the sale.
Another example: I am a photographer uploading my images to a stock image site. As part of ToS I agree to provide a license to host, display, and relicense to buyers on my behalf. The stock site now offers an AI that create new images based on its portfolio. The catch is that all attributed works result in a monetary payment to the artists. When buyers license AI generated works based on my images I get a percentage of the sale. The stock site is legally compliant because it has a license to use my work, and I receive fair compensation when the images are used. The cycle is complete.
It gets trickier in practice, but licensing and compensation is the crux of the matter.
deleted by creator
That’s fine, but not the primary issue.
At some point these companies will need to get licenses for any copyrighted work that was part of the training data, or start over with public domain works only. The art may be data, but that data has legal owners whose rights grant control over it’s use.
Another way to think about is proprietary code. You can see it and learn from it at your leisure. But to use it commercially requires a license, one that clearly defines what can and cannot be done with it, as well as fair compensation.
But it’s not reposting copyrighted images. It is analyzing them, possibly a long time ago, then using complex math and statistics to learn how to make new images when requested, on the fly. It’s an automated version of the way humans learn how to make art or take pictures. If it happens to produce Mario very closely it is because it learned very well.
That is why this isn’t cut and dry. And why it might be good to think of it as derivative works. I don’t think you will be able to nail down this idea of imagination and inspiration. It’s just not that straight forward.
Edit: Also, the generator is not pumping out copyrighted images intentionally. It is waiting for a prompt from a user. Who will then go and post it somewhere. If it is too close to Mario, it is that human user who has violated copyrights. They only used the generator as a tool. I feel like that is very relevant.
That’s the ideal and how it’s advertised, but in reality they retain a lot more of the original copies than most people think, in addition to the fact that the output is often not sufficiently “transformative” in copyright terms to avoid being considered a derivative work still needing a copyright license.
In your Mario example, the character as such is unique enough that it has its own copyright and you can’t use images of that character commercially without a license regardless of how the image was created. If it’s recognizable as Mario then you copied the design as far as a judge would be concerned. If you asked a human to draw it then it would be equally infringing.
A human doesn’t even need to ask for a copyrighted work for it to generate infringing outputs.
In the case of Mario, it’s not literally copyright for the most part, but other IP protections such as trademarks, logos, etc.
I don’t think intention or prompting matters much. If you type “Mario movie” into the YouTube search box and it shows you the Mario movie, YouTube needs to license that material, even if you explicitly ask it to do so and even if you don’t redistribute it. An AI tool is in a similar situation, you still need to license content if you’re making a tool.
I think the problem is that the user often doesn’t realize that they’re infringing on copyright because the prompts don’t specifically mention the copyrighted work.
In other word, there is a need for a way to tell the AI that only public domain work can be used in the generation.
Very possible people don’t realize. And you know what? We shouldn’t care. But if someone generates a Mario and puts in on their website or makes a fanfic comic, doesn’t matter how they made it… go after that person. Just like you always have, Nintendo…
But I worry for the future of any tool if they win this. Add a feature to a computer art tool that feels too “generatey”, you better watch out… I worry about human artists, having to prove the sources they learned from were not protected copyrights when they lean into a style that feels like Nintendo’s…
I’m not an infrengment lawyer… but disney and nintendo, and NYtime and whole lot of artist seems to think they have a case. I suspect it is the same as using a sample from a beyonce song in something you are selling, you may have a problem with beyonce’s jurists
I’m not a lawyer either, but I’m fairly sure that every plaintiff thinks they have a case.