Getty Image is threatening to sue artificial intelligence developer Stability AI for illegally using its photos to train its software.
The American photo agency Getty Images threatened Tuesday to sue the software publisher Stability AI, which used its photos to train an artificial intelligence generating images. Getty distributes illustrative and news photos, including AFP images.
“Stability AI has illegally copied ‘on the internet’ and reprocessed millions of copyrighted images,” Getty said in a statement. The company did not seek to obtain a license adapted to the training of artificial intelligences from Getty, whereas this one exists, regretted the photo agency.
Stability AI is the author of Stable Diffusion, an artificial intelligence that allows to generate images in a given style or inspiration from a simple text.
It is part of the same family of artificial intelligences as 2 of the start-up OpenAI, star of social networks last year for its ability to generate a very wide variety of images on demand. Stability AI, like Dall-E 2, are so-called “generative” artificial intelligences, which produce new images thanks to those presented to them by their author in the training phase.
Stability AI already singled out
Getty Images said it had already begun the first legal steps to take the case to the High Court in London. Stability AI is already the subject of a class action launched in the United States by three artists who also accuse it of violating their reproduction rights.
Artificial intelligences that use large amounts of text or images on the internet to train themselves pose an unprecedented legal problem. Is this a legitimate use of publicly exposed content, as defended by the designers of artificial intelligence, or a violation of the intellectual property rights of the holders of these texts and images?
Stability AI did not respond to requests from AFP.
