Stable Diffusion is one of the most popular generative artificial intelligences in recent times, especially during the second half of 2022, when these models of creating images from text descriptions began to grab headlines and arouse the interest of the general public. The most popular is probably Dall-E 2, which we already tested a few months ago, but it shares the limelight with the aforementioned Stable Diffusion and Midjourney.
It is true, however, that with the arrival of ChatGPT it seems that interest has turned towards text-generating AIs, especially for their future plans, which will soon become part of all kinds of applications and services, from Internet browsers to productivity software. However, over time it is more than likely that new uses will emerge and that, in the end, they will all end up sharing the limelight.
Be that as it may, there is a controversy associated with generative AIs, specifically with all those that have been trained with content protected by copyright. The problem is not, strictly speaking, that these artificial intelligences have been trained with said contents, but rather that, due to their operating system, they have the bad habit of reproducing, in a too literal way, precisely those contents with which they have been previously trained.
Until now we have heard of some cases of individuals who have denounced the companies that create these artificial intelligences, but now we are facing a case of greater depth. And it is that, as we can read in The Verge, Getty Images has sued Stability AI, creators of Stable Diffusion. The reason? You can already imagine it, the popular photography agency and image bank accuses of having used its contents to train Stable Diffusion.
More specifically, Getty Images accuses Stability AI of “Blatant infringement of Getty Images intellectual property on a staggering scale«. It claims that Stability AI copied over 12 million images from its database.”without permission…or compensation…as part of their efforts to build a competitive business“, and that the startup has infringed both the company’s copyright and trademark.
A few days ago we told you that Google has designed a generative AI for music based on text descriptions, but at least for now it has decided not to publish it. We can assume that the main reason has to do precisely with this problem. Since MusicLM has been fed many thousands of hours of music, surely much of it protected by copyright, the search engine company will want to avoid lawsuits such as the one filed by Getty Images against Stable Diffusion.