Skip to Main Content

Research Guides

Artificial Intelligence for Image Research

A guide on how to use Generative AI for image generation, editing, concept creation and development.

Art, Artists, Copyright

Art and artist reactions to AI have been mixed, with many dissenting and valid opinions on the subject and how copyright law should function in regard to it. Currently, lawsuits are in progress to discern if the use of datasets scraped from the internet by AI developers constitutes copyright infringement. 



In order to sufficiently train a diffusion model to create high-quality images, the training dataset must be large enough and of high quality. This leads to the main problem with these models, the sourcing of the training dataset. Popular AI Image Generation models like Midjourney and Stable Diffusion are trained on the LAION dataset, a collection of 5.6 billion images scraped from the internet, including digital artist communities like DeviantArt and ArtStation, along with Getty Images and Pinterest. This has led to a class action lawsuit between a group of artists and the AI research labs, StabilityAI and Midjourney.

The class action suit aims to protect artists by asserting that since the name of the artist is invoked in the text prompt, further AI outputs are considered “derivatives” of that work, and are subject to copyright.

Multiple other lawsuits are also progressing in parallel, such as that from stock photo hosting sites such as Getty Images, which is suing because StabilityAI and Midjourney trained their models on these datasets without consent.

Interpolations of CelebA-HQ 256x256 images with 500 timesteps of diffusion

Jonathan C. Ho, Ajay Jain, Pieter Abbeel


Part of the lawsuit against Stability AI and Midjourney argues that diffusion models can store their training images in a compressed format, known as latent images. It is then found that the images could then be interpolated to produce new derivative images. The argument is that the resulting output image is not a “new” face but formed from pixel-by-pixel interpolation between the two originals, thus constituting a copyright violation.


Artist Opinions:

Graphic Novelist Tomer Hanuka, whose work has been devalued by AI imitations; Mike Winkelmann, who has made millions through quick iterations of daily art by leaning into new technologies; and Jason M. Allen, who augments his work with a mix of AI and digital editing.


3/400 of the Punks by Hanuka collection, created by an anonymous user using AI image generation and sold on OpenSea

“It just doesn't make sense to pay an artist and to wait three weeks if you can get it for essentially nothing.”- Tomer Hanuka

Graphic Novelist and Tomer Hanuka discovered late last year that someone was creating custom models based on his work as NFT’s and selling them on the NFT marketplace OpenSea. Hanuka had no idea who to contact, and the anonymous seller was impossible to track down. This greatly devalued Hanuka’s work by showing how 400 images can be made in his style in a very short amount of time, and with very little effort. This muddies his style, a language he’s been developing for a lifetime, through new (fake) images associated with his name.

Everydays: The First 5000 Days

Mike Winkelmann (Beeple)


Mike Winkelmann, aka Beeple, is a graphic designer who has found commercial success as one of the originators of the “everyday” movement, creating an image from start to finish every day for what is now 10 years. In 2021 Mike was your regular digital artist. In 2023 he is the third-most-expensive living artist having sold his collection of Everydays: The First 5000 Days for $69.3 million, and picking up freelance jobs from musicians and game studios. Mike's lifestyle revolves around constantly sharing and giving away a significant part of himself without any charge and the emergence of AI has provided Winkelmann with a fresh avenue to publish a new digital artwork every day, eliminating tedious tasks and granting him more time for the thoughtful contemplation of creative ideas. He believes that budding digital artists should follow his example and fully utilize the system, embracing it wholeheartedly.


Théâtre d'Opéra Spatial

Jason M. Allen


In 2022, Jason M. Allen won the Colorado State Fair’s annual art competition in the contest for emerging digital artists. One of the first A.I. generated pieces to win such a prize, this contest win caused huge backlash from artists who accused Mr. Allen of cheating. Mr. Allen has openly defended and declared that A.I. was used in the generation of this artwork. Twitter users wrote “We’re watching the death of artistry unfold right before our eyes,”, while some artists defended Mr. Allen, saying that AI is no different than using other digital image manipulation tools like Photoshop or Illustrator, and that prompt engineering itself is a creative endeavor.