GENEVA — The landmark copyright lawsuit filed by NFT artists against generative AI firms took a highly controversial turn on Wednesday, as the legal defense for the AI conglomerates provided a robust counter-argument that challenges the very definition of digital ownership. The defense asserts that the mass scraping of blockchain-registered images constitutes a form of “transformative analysis” that is legally protected under the “fair use” doctrine of international copyright law.
The core of the AI firms’ argument relies on the technical mechanics of how generative models are trained. They assert that their algorithms do not store or replicate individual images; instead, they analyze millions of data points to identify mathematical patterns, styles, and lighting techniques. The resulting AI-generated art, they argue, is a novel creation that bears no direct legal relationship to the specific images utilized in the training set.
This defense creates a profound dilemma for the NFT sector. If the court rules in favor of the AI firms, it effectively creates a legal loophole where digital scarcity can be mathematically analyzed and replicated by an autonomous algorithm without compensation to the original creator. Conversely, a ruling in favor of the artists could impose massive, perhaps insurmountable, liability on the entire AI development industry.
“This case is the Battle of the Waterloo for digital property rights,” a prominent technology lawyer observed from Geneva. “The court must decide if a cryptographic token provides an artist with the right to exclude an algorithm from ‘learning’ from their work. The outcome will definitively determine whether the blockchain can protect human creativity from being effectively commoditized by artificial intelligence.”
calling mass scraping of artists work transformative analysis is such a stretch. they literally need the training data to function, that IS copying
they need the training data to function. calling that transformative analysis is like saying a thief transformed your TV by watching it
transformative analysis? they just stole the data to train AI
the fair use defense basically says its not copying because we transformed it into math. thats a convenient loophole for a trillion dollar industry
Elif Demir transforming it into math is such a convenient defense. the math literally cannot exist without the original training data
Elif Demir’s fair use defense is just saying theft is fine if you transform it into math. Convenient for a trillion dollar industry
Elif Demir’s fair use defense is just saying theft is fine if you transform it into math. Convenient for a trillion dollar industry
mass scraping as transformative analysis is such a joke
the problem is if the artists win here, it could kill the entire AI industry. courts hate rulings with that kind of collateral damage
^ that doesnt make it right. you cant just say ‘well we built a huge business on stolen data so we should be allowed to keep doing it’
Kofi M. the collateral damage argument is exactly why courts punt on these rulings. look at the oracle v google precedent, took a decade
artists winning could kill AI development but AI winning kills creative ownership. theres no good outcome here, just less bad ones
Ava Chen-Ramirez nailed the dilemma. artists win = AI dies, AI wins = ownership dies. theres no clean outcome here
Kofi M has a point but collateral damage doesn’t make scraping right. Transformative analysis is a joke
Kofi M has a point but collateral damage doesn’t make scraping right. Transformative analysis is a joke
Kofi M. the collateral damage argument cuts both ways. killing creative ownership to preserve AI is also catastrophic collateral damage
blockchain-registered images being scraped to train models that then compete with those same artists is peak irony. hope the court sees through the fair use argument
blockchain images scraped to train models that compete with original artists
AI firms calling mass scraping transformative analysis is like a bakery saying eating all the flour is research. the output literally competes with the original artists
if the court sides with AI firms on fair use it destroys the entire premise of NFT scarcity. your 6 ETH piece becomes worthless when an algorithm replicates the style for pennies
mira the NFT premise was always shaky though. tokenizing a JPEG doesnt prevent anyone from copying the image. copyright court was always going to be the real battle
calling it transformative analysis when your model literally cannot function without the training data is peak legal gymnastics
training_data_refugee the models dont store images argument is such cope. they encode style and composition precisely enough to reproduce near-identical outputs. thats reproduction with extra steps
training_data_refugee the models dont store images argument is such cope. they encode style and composition precisely enough to reproduce near-identical outputs. thats reproduction with extra steps
if the court rules fair use for training on blockchain registered images, NFT scarcity becomes mathematically meaningless. anyone can train a model on your collection and output derivatives
if the court rules fair use for training on blockchain registered images, NFT scarcity becomes mathematically meaningless. anyone can train a model on your collection and output derivatives