From MIT Technology Review:
Those cool AI-generated images you’ve seen all over the internet? They are most likely based on the works of Greg Rutkowski.
Rutkowski is a Polish digital artist who uses classical painting styles to create dreamlike fantasy landscapes. He has done artwork for games like Sony’s Horizon Forbidden West, Ubisoft’s Anno, Dungeons & Dragons, and Magic: The Gathering. And he has become a sudden success in the new world of text-to-image AI generation.
His distinctive style is now one of the most used messages in the new open source AI art generator Stable Diffusion, which launched late last month. The tool, along with other popular image generation AI models, allows anyone to create stunning images based on text messages.
For example, type “Wizard with sword and glowing orb of magical fire fights a ferocious dragon Greg Rutkowski,” and the system will produce something that doesn’t look a million miles from Rutkowski-style works.
But these open source programs are created by pulling images from the internet, often without proper permission and attribution to the artists. As a result, they are raising tough questions about ethics and copyright. And artists like Rutkowski have had enough.
According to the Lexica website, which tracks more than 10 million images and ads generated by Stable Diffusion, Rutkowski’s name has been used as an ad some 93,000 times. Some of the most famous artists in the world, such as Michelangelo, Pablo Picasso, and Leonardo da Vinci, listed around 2,000 indications each or less. Rutkowski’s name also appears as a message thousands of times on Discord from another text-to-image generator, Midjourney.
Rutkowski was initially surprised, but thought it could be a good way to reach new audiences. He then tried to look up his name to see if an article he had worked on had been published. The online search found work that had his name attached to it but was not his.
“It’s only been a month. How about in a year? I probably won’t be able to find my job out there because [the internet] it will be inundated with the art of AI,” says Rutkowski. “That is worrisome”.
Stability.AI, the company that built Stable Diffusion, trained the model on the LAION-5B dataset, which was compiled by the German nonprofit organization LAION. LAION assembled the dataset and narrowed it down by filtering out watermarked images and unaesthetic ones, such as logo images, says Andy Baio, a technologist and writer who downloaded and analyzed some of the Stable Diffusion data. Baio analyzed 12 million of the 600 million images used to train the model and found that a large portion of them came from third-party websites like Pinterest and art-buying sites like Fine Art America.
Many of Rutkowski’s artworks have been pulled from ArtStation, a website where many artists upload their portfolios online. Its popularity as an AI indicator stems from several reasons.
First, her fantastical and ethereal style looks great. He is also prolific and many of his illustrations are available online in high enough quality that there are plenty of examples to choose from. An early text-to-image generator called Disco Diffusion offered Rutkowski as an example.
Rutkowski also added alt text in English when uploading his work online. These image descriptions are useful for visually impaired people using screen reading software and also help search engines rank images. This also makes them easy to scratch, and the AI model knows which images are relevant to the prompts.
. . . .
Other artists besides Rutkowski have been surprised by the apparent popularity of his work on text-to-image generators, and some are now fighting back. Karla Ortiz, a San Francisco-based illustrator who found her work in the Stable Diffusion dataset, has been raising awareness about AI art and copyright issues.
Artists say they risk losing revenue as people start using AI-generated images based on copyrighted material for commercial purposes. But it’s also much more personal, Ortiz says, arguing that because art is so closely tied to a person, it could raise privacy and data protection issues.
“There is a coalition growing within the artist industries to figure out how to address or mitigate this,” says Ortiz. The group is in its early days of mobilization, which could mean pushing for new policies or regulations.
Link to the rest at MIT Technology Review and thanks to R. for the suggestion in the comments.
PG predicts that more than one copyright infringement lawsuit will be filed against various individuals, institutions, and companies that provide AI services in which an artist’s copyrighted work was used to seed the AI.
In the United States, such lawsuits will almost certainly be filed in the federal court system, since copyright is governed by federal law. Some states have laws that appear to grant exclusive rights to publish state documents to the state or to those to whom the state has given permission to make and sell copies of state documents, but attempting to protect a creative work from republication in any other way than not is in compliance with federal copyright laws and rulings is generally considered nonsense.
One thing judges do when faced with a novel question is to build on similar situations that have occurred before.
As a crude example, if an individual uses a computer and a software program created by a third party to make an exact copy of the text of a copyrighted book, the manufacturer of the computer or the company that created and sold the word processor The program used to make a copy of the book will not be held responsible for copyright infringement because it only provided tools and the author used the tools in the way he chose.
AI art programs require a notice to create any image. The OP mentions using an artist’s name in an AI notice as a way to generate an image.
However, that decision is not made by the creators/owners of the AI program, but by the user. The creators of the AI program ran a large number of images from a large number of artists through the program’s processor. Is it a violation of copyright law to link an artist’s name to a painting the artist created? PG doesn’t think so.
In fact, using Mr. Rutkowski’s work without attribution would also be an offense against the creator.
PG does not consider the creation of works “inspired by” an artist to constitute copyright infringement when they are not copies of what the artist created or closely resemble what an artist created. PG does not believe that an artistic style is protected by copyright.
If PG’s understanding of the way AI art programs work is to deconstruct the original copy of the image into its component parts and assign some kind of marker to the parts, so that a message for a big building à la of the British Museum do not. generate an image dominated by a dragon.
PG just created an ad, “Windsor Castle sailing on the ocean” and ran it through an AI art generator. This is what he got.
PG then modified their ad to read “Greg Rutkowski Ocean Sailing Windsor Castle” and this is what he got:
For one last experiment, PG created another ad with a different artist, “Windsor Castle Sailing On The Ocean Andy Warhol” and this is what came up.
PG is no art expert, but he doesn’t think any of his AI illustrations will break Rutkowski or Warhol.