This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:


reputable news agency


'Glaze' software aims to protect art from copycat AI

Credit: CC0 Public Domain

Researchers at the University of Chicago have released free "Glaze" software that they say can thwart efforts by generative artificial intelligence (AI) to copy an artist's style.

The program makes tiny changes to that, while invisible to , act as a " cloak" when they are posted online, the team behind the explained on their website.

If generative AI finds a Glaze-guarded image online, it is prevented from correctly analyzing and copying the style, the team said.

Glaze was created at the behest of artists outraged that programs like Midjourney and Stable Diffusion, schooled on troves of images available online, could mimic their styles on command.

"AI has been evolving too fast, and there must be some guardrails or regulations around it," said Shawn Shan, the doctoral student in charge of the project.

"The goal of this is to push back from a technical standpoint."

The team behind Glaze worked with artists including the illustrator Karla Ortiz, who is among the plaintiffs in a US court case against several firms with image-producing generative AI services.

"If Karla uses our tool to cloak her artwork, by adding tiny changes before posting them on her online portfolio, then Stable Diffusion will not learn Karla's artistic style," the lab team said.

"Instead, the model will interpret her art as a different style—for example, that of Vincent van Gogh."

The creators of Glaze concede it is not a panacea, given how quickly AI evolves.

The hope, the team said, is that Glaze and similar projects will protect artists at least until defensive laws or regulations can be implemented.

Glaze has been available for free download since March 15.

© 2023 AFP

Citation: 'Glaze' software aims to protect art from copycat AI (2023, March 27) retrieved 30 May 2024 from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Explore further

Is there a way to pay content creators whose work is used to train AI? Yes, but it's not foolproof


Feedback to editors