General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsInvestors sue Adobe execs over AI copyright statements
https://courthousenews.com/investors-sue-adobe-execs-over-ai-copyright-statements/-snip-
Additionally, they say the defendants including former Adobe Chief Executive Officer Shantanu Narayen; Adobe director Christiano Amon, the current president and CEO of Qualcomm Incorporated, a global wireless tech company; Claude Alexandre, Adobes vice president of digital media; and over ten other high-ranking employees at Adobe signed and approved statements that Adobe didnt infringe any copyrighted material as part of training its AI software.
-snip-
According to the stockholders, one of Adobes AI training models relied on a dataset that consisted of approximately 196,640 books copied from Bibliotik, an online shadow-library known for distributing copyrighted literary works without authorization from authors, publishers, or other rights holder.
-snip-
Adobes officers and directors knowingly allowed the company to build its AI strategy on the unlawful infringement of copyrighted materials, the plaintiffs attorney Frank Bottini said in a statement to Courthouse News. This shareholder derivative suit seeks to hold Adobes top leadership liable for destroying billions of dollars in shareholder value.
-snip-
The entire generative AI industry is built on the theft of copyrighted intellectual property.
AZJonnie
(4,193 posts)There's been 1000's of cases just like this one, ever since "stocks" were a thing. Company execs misled shareholders about some *thing*, and those lies cost the stockholders a bunch of money.
In the end I'd guess it'll hinge mostly on whether or not it can be proven that the managers knew, or should have know, they were misleading investors. Discovery (emails and such) could be a real thorn for these execs
In the end, who wins will probably come down to whether Adobe execs said or signed of on statements that were categorically and unambiguously claiming the company "are not infringing on copyrights while training its AI". If they really did that, the plaintiffs I'm guessing will win, esp. since that site is well-known for pirated materials.
However if what they actually did was "characterizing the strategy as a generational opportunity to enhance productivity and creativity through responsible, commercially safe models natively integrated into its software ecosystem, it might be a tougher slog for plaintiffs. Because the key words here don't have hard definitions. They're weasel words and marketing-speak.
I don't know either way what the truth is (though I suspect the plaintiffs have 'em dead to rights), but that'll be the basic question.
Another interesting piece of this is that I would bet AI generally is playing havoc on Adobe's business model. "Why do I need Photoshop if I can just tell AI to conjure up the image I want" type of thing. So it may turn out legally difficult for shareholders to positively attribute the companies recent loss in value to their AI training scandal. That might not effect whether they win generally, but could have a large effect on how much they receive in recompense.
Another interesting twist would be if it turned out that the AI companies stole Adobe code to train their models to do what they do
I'll be watching with some interest, hope you'll keep us posted HPD, and keep their feet to the fire!
highplainsdem
(63,693 posts)property.
This lawsuit is particularly interesting to me because it highlights that image/video generators ripped off text as well.
The same goes for music generators. Suno and Udio's theft of all the copyrighted music they could access always gets a lot of attention, but no one should overlook the fact that the text prompts used to generate music work ONLY because so much text was stolen that describes that music.
And you can bet that the text Adobe stole included all the writing about art, photos and film/video they could get, so the AI would be able to link users' prompts to some approximation of what was being requested.
As I've said before, I'd like to see the AI companies sued out of existence, and the people in those industries responsible for the IP theft prosecuted and ideally sentenced to some time in prison.
Having them sued by investors they lied to wouldn't be quite as satisfying as seeing them sued by the people whose work they stole (the original sin of genAI, where the dishonesty started), but investors have more money for suing them.
AZJonnie
(4,193 posts)With significant penalties for past sins attached paid to rights holders, and with a strictly enforced licensing scheme in place going forward. Make them pay for the use of the training materials. There's plenty of huge companies with huge catalogs (publishing houses) who I'm sure would happily sell training rights to the AI companies (hell, they're just being ripped off now, better to make some dough instead). There's not really anything about how it all works that firmly necessitates work being "stolen". They stole it cause they didn't want to pay if they didn't have to. Make it so they HAVE TO. If that puts them out of business, so be it.
Three problems with that though is that only the really big players will able to afford to pay those fees, and the fact that unless it's done and enforced worldwide, these companies will just move, and countries like China probably won't make the same efforts to honor the system, which will put them in a commanding position in the marketplace.