Glaze
In plain wordsA free tool that lets artists treat their pictures before posting them, with changes people barely notice, so that an AI trained to imitate their style learns a different one.
A tool that applies a small, optimized perturbation (a 'cloak') to an artwork so that the feature extractors used by text-to-image modelstext-to-image model, n.: An AI system that creates a picture from a written description, such as 'a fox reading a newspaper, in watercolor', usually within seconds. perceive it as being in a different style, protecting artists against style mimicry by models fine-tuned on their work. Unlike NightshadeNightshade, prop. n.: A free tool for artists that subtly alters their images so that AI image generators trained on them without permission learn the wrong things — associating 'dog' with pictures of cats, say., it is defensive: it aims to stop imitation of an individual's images rather than to damage models trained on them. Its protection depends on the perturbation surviving image processing and on which models imitators use; researchers have shown that simple purification such as upscaling or adding noise can weaken it, and the tool has been revised in response. Like other adversarial perturbations, it offers no guarantee against future models.
She glazed her new illustrations before uploading them to her portfolio site.
Origin
Developed by Shawn Shan, Jenna Cryan, Emily Wenger, Haitao Zheng, Rana Hanocka and Ben Zhao at the University of Chicago, released in March 2023 and presented at USENIX Security 2023. Robert Hönig and colleagues (2024) showed that low-effort countermeasures could defeat it and similar protections.
Related
- NightshadeNightshade, prop. n.: A free tool for artists that subtly alters their images so that AI image generators trained on them without permission learn the wrong things — associating 'dog' with pictures of cats, say.
- adversarial exampleadversarial example, n.: An input altered on purpose, often too slightly for people to notice, so that an AI model gets it badly wrong — like a panda photo that a classifier confidently calls a gibbon.
- text-to-image modeltext-to-image model, n.: An AI system that creates a picture from a written description, such as 'a fox reading a newspaper, in watercolor', usually within seconds.
- AI copyright litigationAI copyright litigation, n.: The many lawsuits in which authors, artists, news organizations, record labels and others accuse AI companies of copying their work without permission to build or run AI models.
Sources
- Shan, S., Cryan, J., Wenger, E., Zheng, H., Hanocka, R. & Zhao, B. Y. (2023). Glaze: Protecting Artists from Style Mimicry by Text-to-Image Models. USENIX Security 2023. arXiv:2302.04222.
- Hönig, R., Rando, J., Carlini, N. & Tramèr, F. (2025). Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI. ICLR 2025.
Mentioned in
NightshadeNightshade, prop. n.: A free tool for artists that subtly alters their images so that AI image generators trained on them without permission learn the wrong things — associating 'dog' with pictures of cats, say.