Database of 16,000+ artists used to train Midjourney
Database of 16,000+ artists used to train Midjourney
Occurred: February 2022
Report incident ๐ฅ | Improve page ๐ | Access database ๐ข
A database listing the names of over 16,000 artists was purportedly used to train the Midjourney image generator, prompting an outcry.
Including names such as Banksy, David Hockney, Frida Kahlo, Yayoi Kusama and Damian Hirst, the 'Midjourney Style List' was allegedly used during a process of refining the model's ability to mimic works of the selected artists and their styles. These outputs were then prominently featured as reference material for image creation.
The list was first published to a Discord server in February 2022 by Midjourney CEO David Holz, who welcomed the addition of the artists' names to the training of the model.ย
Part of the list was included in a court document (pdf) filed late November 2023 as part of a class-action lawsuit against DeviantArt, Midjourney, Stability AI and Runway AI.
These models made use of LAION-5B, a nonprofit, publicly available database that indexes more than five billion images from across the Internet, including the work of many artists.
The emergence of the list raised questions about possible copyight violations by Midjourney and the other named entities. It also reinvigorated a broader debate about copyright and consent in the generation of AI images.
Operator: ย
Developer: Midjourney
Country: USA
Sector: Media/entertainment/sports/arts
Purpose: Train model
Technology: Database; Machine learning
Issue: Cheating/plagiarism; Copyright; Ethics/values; Transparency
https://www.artnews.com/art-news/news/midjourney-ai-artists-database-1234691955/
https://www.theartnewspaper.com/2024/01/04/leaked-names-of-16000-artists-used-to-train-midjourney-ai
https://www.nbcnews.com/tech/tech-news/famous-artists-trained-ai-generator-viral-list-rcna131995
https://hyperallergic.com/864947/database-of-artists-used-to-train-ai-leaks-to-the-public/
Page info
Type: Incident
Published: January 2024