About
Glaze is a free tool developed by researchers at the University of Chicago's SAND Lab, designed to protect artists from having their unique styles scraped and mimicked by AI image generation models. It works by applying subtle, mathematically calculated pixel-level perturbations — called 'cloaks' — to artwork images before they are posted online. These changes are nearly imperceptible to the human eye, but cause AI training pipelines to misinterpret the underlying style, associating the artwork with an entirely different aesthetic. When an AI model is trained on glazed images, its attempts to replicate that artist's style produce distorted or inaccurate results, effectively breaking the style-copying pipeline. All image processing is performed locally on the user's machine, meaning artwork is never uploaded to external servers — a significant privacy advantage for artists concerned about their work leaving their control. Glaze offers adjustable cloaking intensity levels, allowing artists to balance protection strength against the visibility of changes in their images. It is available as a desktop application for both Windows and macOS, as well as a web-based version for users who cannot run the local app. The tool is free and aimed at illustrators, digital artists, and any creator who shares work online and wants to guard against unauthorized AI style replication.
Key Features
- Style Cloaking: Applies imperceptible pixel-level perturbations to images that mislead AI training models into associating the artwork with a completely different artistic style.
- Local Processing: All image cloaking is performed on the user's own machine — artwork is never uploaded to external servers, preserving full privacy and control.
- Adjustable Protection Intensity: Artists can select from multiple cloaking strength levels, trading off between stronger AI protection and the degree of visible change to the image.
- Cross-Platform Availability: Available as a native desktop app for Windows and macOS, as well as a browser-based web version for broader accessibility.
- Research-Backed Approach: Developed by the SAND Lab at the University of Chicago, with peer-reviewed research validating its effectiveness against major generative AI model training pipelines.
Pros
- Completely Free: Glaze is an academic research project offered at no cost, with no subscription tiers or paywalls for any of its features.
- Privacy-First Design: Local processing ensures artists' original artwork never leaves their device, eliminating the risk of exposure to third-party servers.
- Minimal Visual Impact: Cloaking perturbations are designed to be nearly invisible to the human eye at normal viewing distances, preserving the intended appearance of the artwork.
Cons
- Not Foolproof: As AI training techniques evolve, future model architectures may reduce the effectiveness of current cloaking methods, requiring ongoing updates.
- Processing Time: Generating cloaked images can be computationally intensive, particularly on older hardware or for high-resolution artwork, leading to longer wait times.
- Retroactive Protection Not Possible: Glaze only protects images before they are posted; artwork already scraped and used in AI training datasets cannot be retroactively protected.
