Hayao Miyazaki, Matt Groening and Keith Haring among artists allegedly used to train Midjourney AI

A database listing thousands of artist names has been allegedly compiled by Midjourney developers to train the text-to-image tool.

Date
5 January 2024

Amidst an ongoing class-action lawsuit between a group of artists and major AI companies, it has been revealed that Midjourney gathered a database of artists to train its text-to-image tool, allowing the easy reproduction of hugely popular artistic styles.

A leaked Google Sheet was circulating online on 31 December 2023 featuring visual styles like “xmaspunk”, “carpetpunk” and “polaroidcore” and thousands of individual artist names. The list was allegedly put together by developers at Midjourney; screenshots shared by Riot Games’ Jon Lam (founder of advocacy moment Create Don’t Scrape) appeared to show developers discussing sourcing artists and styles from Wikiart and laundering datasets to avoid copyright infringement.

The list spans huge names from history as well as some of the biggest creatives in visual culture today, with artists like William Blake and Henri Matisse gathered alongside the likes of Keith Haring, Matt Groening, Wes Anderson, Katsuhiro Otomo, Frida Kahlo, Hayao Miyazaki, Banksy, David Hockney, Trove Jansson, Anish Kapoor and Yayoi Kusama. Developers even discussed emulating the style of Bob Ross in internal conversations.

Not only leading names have been pooled by Midjourney – one X user (previously Twitter) pointed out that six-year-old Hyan Tran was on the database, who contributed artwork for a 2021 charity trading card edition of Magic: The Gathering raising money for Seattle Children’s Hospital Autism Center.

The original Google Sheet is no longer available to view but another version has been uploaded on the Internet Archive. It was first published as evidence in the ongoing legal battle between several artists and Midjourney, Stability AI and DeviantArt – a judge dismissed claims against Midjourney and DeviantArt in October. The document was taken from 455 pages of supplementary evidence filed in November.

Artists who have had their work scraped by AI models are facing uncertain legal territory when it comes to AI and copyright, especially as it becomes increasingly difficult to show substantial similarity between an artist’s work and AI outputs when models interpret elements of an artist’s style, like linework, rather than creating identical copies.

Hero Header

Keith Haring: Tuttomondo (All World). Paolo Gallo – stock.adobe.com

Share Article

About the Author

Liz Gorny

Liz (she/they) joined It’s Nice That as news writer in December 2021. After graduating from the University of Bristol, they worked freelance, writing for independent publications such as Little White Lies, Indie magazine and design studio Evermade.

It's Nice That Newsletters

Fancy a bit of It's Nice That in your inbox? Sign up to our newsletters and we'll keep you in the loop with everything good going on in the creative world.