{"id":72633,"date":"2023-01-17T14:46:05","date_gmt":"2023-01-17T12:46:05","guid":{"rendered":"https:\/\/forklog.com\/en\/?p=72633"},"modified":"2025-09-09T09:28:48","modified_gmt":"2025-09-09T06:28:48","slug":"artists-sue-stability-ai-and-midjourney-over-training-their-image-generation-algorithms-on-their-images","status":"publish","type":"post","link":"https:\/\/u1f987.com\/en\/artists-sue-stability-ai-and-midjourney-over-training-their-image-generation-algorithms-on-their-images\/","title":{"rendered":"Artists sue Stability AI and Midjourney over training their image-generation algorithms on their images"},"content":{"rendered":"

Three artists have filed a class-action lawsuit against the developers of image-generating AIs over alleged copyright violations. This is reported by The Register<\/a>.<\/p>\n

The case concerns Stability AI, Deviant Art and Midjourney. The plaintiffs allege that the developers used protected images to train their algorithms.<\/p>\n

\n

“Stable Diffusion contains unauthorized copies of millions \u2014 and possibly billions \u2014 of images protected by copyright. They were made without the knowledge or consent of the artists,” said attorney Matthew Butterick.<\/p>\n<\/blockquote>\n

In August 2022, Stability AI released the text-to-image tool Stable Diffusion. It was trained on a pair of “image-caption” pairs sourced from the internet. The tool stores compressed copies of training data and uses them to recombine into a new “collage” based on existing images created by people.<\/p>\n

According to Butterick, the resemblance between the final images and the training data does not matter, because they compete with the originals in the market.<\/p>\n

\n

“At a minimum, Stable Diffusion’s ability to flood the market with an almost unlimited number of drawings infringing copyright will inflict irreparable damage on the art market and on artists,” he said.<\/p>\n<\/blockquote>\n

The paid app DreamUp, developed by DeviantArt, was also built using Stable Diffusion. Midjourney, in turn, offers a text-to-image generator trained on pictures downloaded from the internet.<\/p>\n

The three plaintiffs include artists Sara Andersen, Kelly McKernan and Carla Ortiz. They seek damages and injunctions on behalf of all artists harmed by Stable Diffusion.<\/p>\n

\n

“People cannot help bringing their humanity to art. Art is deeply personal, and AI has just stripped humanity from it, reducing a lifetime of work to an algorithm,” Andersen said.<\/p>\n<\/blockquote>\n

McKernan said that her name had been used at least 12,000 times to generate images in Midjourney. According to the artist, she has earned nothing from it.<\/p>\n

\n

A reminder that my name has been used and profited from at a MINIMUM of 12k times. These are just the publicly available results. Many have private servers they pay to keep. What if I had been paid for every instance? That\u2019s life changing money for a single mom. This isn\u2019t right. pic.twitter.com\/6wvnnE4NRO<\/a><\/p>\n

\u2014 Kelly McKernan (@Kelly_McKernan) January 15, 2023<\/a><\/p><\/blockquote>\n