Accurate, Focused Research on Law, Technology and Knowledge Discovery Since 2002

UChicago scientists develop new tool to protect artists from AI mimicry

University of Chicago News: “A new tool allows artists to upload digital images with slight changes that are nearly invisible to the human eye, but confound AI art generators searching for references. “Glaze” program makes miniscule changes to confound AI art generators. Last year, the arrival of powerful AI models capable of generating original images based on text descriptions amazed people with their uncanny ability to simulate different artistic styles. But for artists, that wonder was a nightmare, with the easy, startlingly accurate mimicry threatening their livelihood. Now, a team of University of Chicago computer scientists have built a tool that protects artists from the absorption of their style into these AI models. Called Glaze, the software “cloaks” images so that models incorrectly learn the unique features that define an artist’s style, thwarting subsequent efforts to generate artificial plagiarisms. The research, developed by the SAND Lab research group led by Neubauer Professors of Computer Science Ben Zhao and Heather Zheng, gives artists a countermeasure against generative art platforms such as DALL-E, Midjourney, and Stable Diffusion, which have exploded in popularity…”

Sorry, comments are closed for this post.