What is Glaze? How to use it to protect my art from AI scraping?

Friendly Neighborhood Artist
10 Jul 202310:43

TLDRThe video discusses 'Glaze,' a tool designed to protect artists' work from being used in AI models, specifically those utilizing stable diffusion for fine-tuning styles. Glaze prevents new pieces from being trained on but cannot protect existing art already in AI systems. It is not a foolproof solution, yet it has proven effective against attempts to break it. The process involves adding an invisible layer to artwork that is detectable only by AI, without altering the human visual experience. The video also explores the impact of different settings on the Glaze process, which can introduce visible artifacts into the art, thus requiring fine-tuning for each piece. The artist demonstrates the Glaze application and shares their findings on the effectiveness and visual trade-offs of using Glaze to safeguard their work against AI scraping.


  • 🎨 **Glaze is a tool for artists**: It's designed to protect artwork from being used in AI models, specifically in style transfer techniques known as 'stable diffusion'.
  • 🚫 **Prevents new art from AI training**: While it can't protect existing art that's already been incorporated into AI systems, Glaze can prevent new pieces from being used for AI training.
  • πŸ” **Invisible to the human eye**: The modifications made by Glaze to artwork are not visible to humans but are detectable by AI models, effectively blocking them from using the artwork.
  • πŸ›‘οΈ **Current protection status**: As of the knowledge cutoff in March, no countermeasures have successfully broken through Glaze's protection against AI training.
  • πŸ–ΌοΈ **Examples of Glazed Art**: The script provides examples of original and Glazed artwork, with minimal visible differences to the human eye.
  • πŸ”§ **Adjustable settings**: Glaze allows artists to adjust the magnitude of changes added to their art, which can affect the level of protection and visibility of the changes.
  • ⏱️ **Time-consuming process**: Applying Glaze to artwork can be a lengthy process, with higher settings potentially taking longer to process.
  • πŸ“ˆ **Quality and intensity trade-off**: There's a trade-off between the quality of the final image and the intensity of the Glaze effect; higher intensity provides better protection but may distort the artwork more.
  • 🌐 **Open source and international**: Glaze is an open-source tool developed by students from the U.S. and China, aiming to protect artists' works from unauthorized AI use.
  • πŸ“ **Artistic consideration**: Artists using Glaze may need to inform their audience about the use of the tool to manage expectations about the artwork's appearance.
  • βš™οΈ **Ongoing development**: The tool is still being improved, with developers trying different methods to enhance its effectiveness without compromising the artwork's aesthetic quality.

Q & A

  • What is the purpose of Glaze in the context of art?

    -Glaze is a tool designed to protect artists' work from being used in AI models, specifically to prevent their art from being incorporated into styles for AI image generation processes like stable diffusion.

  • How does Glaze protect an artist's work?

    -Glaze applies a cloak to the artwork that is invisible to the human eye but detectable by AI models. This cloak prevents the AI from using the artwork for training or style generation purposes.

  • Is there a difference between the original and the Glazed version of an artwork?

    -The Glaze process does not significantly alter the appearance of the artwork to the human eye. However, it does introduce changes that are detectable by AI, which can sometimes result in minor visible artifacts depending on the settings used.

  • Can Glaze be used to protect artwork that has already been used in AI models?

    -No, Glaze cannot protect artwork that has already been incorporated into AI models. It is designed to protect new pieces from being used in future AI training or style generation.

  • What are the potential drawbacks of using Glaze on artwork?

    -While Glaze is intended to be invisible to the human eye, using higher settings can introduce visible changes or artifacts to the artwork, potentially affecting its aesthetic quality.

  • How does the Glaze process affect the style of artwork when used in AI models?

    -If an artist's style has already been trained in an AI model, applying Glaze over time can alter the style when people look it up, potentially leading to a transformed style akin to Impressionism.

  • What is the guided filter and why might it not be safe to use with Glaze?

    -The guided filter is a method mentioned in the context of trying to get around Glaze's protection. It is not considered safe because the guidance itself may contain adverse noise, which could potentially be reintroduced by the filter.

  • How does the Glaze tool differ from other tools that claim to disrupt image-to-image attacks?

    -Glaze is an open-source tool that is reportedly five times stronger than other tools in disrupting image-to-image attacks. It is specifically designed to protect against AI training on the artwork it is applied to.

  • What are the settings in the Glaze tool and how do they affect the artwork?

    -The Glaze tool has settings that control the magnitude of changes added to the artwork. Higher values lead to more visible changes but offer stronger protection against AI art training. The artist needs to fine-tune these settings based on each piece of artwork.

  • How long does the Glaze process take and how does it affect the user's computer performance?

    -The Glaze process can take several minutes to over half an hour depending on the settings chosen. It can also significantly slow down the user's computer during the process.

  • What advice does the artist provide for those concerned about their artwork being used in AI training?

    -For those seriously concerned about their artwork being used in AI training, the artist suggests using higher settings on the Glaze tool. For those seeking a normal level of protection, the artist indicates that some level of distortion to the art is to be expected.



🎨 Introduction to Glades for Art Protection

The video begins with an introduction to Glades, a tool designed to protect artists' work from being used in 'Laura,' a fine-tuned style creation process. The speaker discusses their research into stable diffusion, particularly focusing on counteracting Laura's use of art. Glades is presented as a current solution to prevent new pieces from being utilized in this manner, though it does not protect older works already integrated into systems. The speaker also mentions that despite attempts to circumvent Glades, it remains effective, with only AI models being able to detect the subtle alterations it makes to artwork.


πŸ–ŒοΈ Applying Glades: A Practical Demonstration

The speaker proceeds to demonstrate the application of Glades on an artwork. They discuss the process, including selecting the quality of rendering and the magnitude of changes that Glades will introduce to the artwork. The speaker experiments with different settings, noting that higher settings provide stronger protection but may also introduce more visible changes to the art. They share their observations on the effects of Glades at various intensity levels, from high, which significantly alters the image, to lower settings that are more subtle but still affect the artwork's appearance. The speaker concludes that the optimal setting may vary based on the specific piece of art and that artists may need to inform their audience about the use of Glades to avoid misunderstandings about the artwork's quality.


πŸ“’ Conclusion and Community Sharing

The video concludes with the speaker expressing gratitude to the viewers and mentioning their intention to share the processed versions of their artwork on the community tab for closer inspection. They summarize their findings, emphasizing that while Glades can provide a level of protection, it does have the side effect of distorting the art to some degree. The speaker acknowledges that the tool may not be perfect but is currently the best option available for artists concerned about their work being used in AI training. They encourage viewers to continue experimenting with Glades to find a satisfactory balance between protection and visual quality.




Glaze refers to a technique or tool used to protect digital artwork from being utilized by AI systems, specifically in the context of style transfer or training AI models. In the video, it is presented as a method to safeguard an artist's work from being scraped and used to create new styles without permission. The term 'Glaze' is central to the video's theme as it directly addresses the concern of AI infringing on artistic integrity.

πŸ’‘Stable Diffusion

Stable Diffusion is mentioned as a platform where artists' styles can be applied to create new images. It is a type of AI model that uses existing styles to generate images. In the video, the concern is raised that without protection like Glaze, artists' styles could be uploaded to Stable Diffusion and used to create derivative works, potentially without the artist's consent.

πŸ’‘AI Scraping

AI Scraping is the process of extracting data from digital sources, in this case, specifically referring to the unauthorized use of an artist's work by AI systems. The video discusses how Glaze can prevent an artist's work from being scraped and used in AI models, which is a significant concern for digital artists who wish to maintain control over their creations.


Countermeasures in the context of the video are strategies or techniques used to thwart or protect against unwanted actions, such as AI scraping of artwork. The speaker mentions that Glaze has some countermeasures against it, indicating that there are ongoing efforts to improve its protective capabilities against AI models that might attempt to bypass its safeguards.

πŸ’‘Guided Filter

A Guided Filter is a technique in image processing that can be used to modify images in a manner that is guided or influenced by another image. In the video, it is mentioned as a potential method that could be used to circumvent the protection offered by Glazing, although it is noted that this approach may not be safe due to the introduction of adverse noise.

πŸ’‘Control Net

Control Net is referenced as the work of an individual who is trying to find ways around the Glaze protection. It implies a tool or project aimed at controlling or managing the application of AI to artistic works. The mention of Control Net in the video underscores the ongoing battle between protecting artistic work and the advancement of AI technologies.

πŸ’‘Image-to-Image Attacks

Image-to-Image Attacks are a type of AI manipulation where an original image is altered or modified based on textual input, such as adding a hat to a person in the image. The video discusses how Glaze can offer some level of protection against such attacks, preventing AI models from easily making these modifications using an artist's original work.

πŸ’‘Open Source Tool

An Open Source Tool is software whose source code is made available to the public, allowing anyone to view, use, modify, and distribute it. The video mentions that Glaze is an open-source tool created by students from the U.S. and China, emphasizing its collaborative and transparent nature, which allows for community involvement in its development and improvement.


Impressionism is an art movement characterized by the use of small, thin strokes of color to capture the fleeting visual impressions of a scene. In the video, it is used as an example of how an artist's style might be altered if their work is used in AI models without Glaze protection, suggesting that the original style could be distorted to resemble something akin to Impressionism.

πŸ’‘Rendering Quality

Rendering Quality refers to the level of detail and visual fidelity in the final output of an image or video. The video discusses the option to select different rendering qualities when using Glaze, which affects the processing time and the degree to which the artwork is altered to protect it from AI scraping.


Artifacts, in the context of the video, refer to unintended visual elements or distortions that appear in an image as a result of the Glazing process. The artist expresses concern about the artifacts introduced at different settings, as they can detract from the original quality and aesthetic of the artwork.


Glaze is a method to protect art from being used in AI models, specifically in style transfer processes.

AI models like Stable Diffusion can generate styles based on artists' works, which Glaze aims to prevent.

Glaze cannot protect art that has already been integrated into AI systems, only new pieces.

The protection provided by Glaze is not permanent but is currently the only solution available.

Glaze works by embedding a cloak within the image that is invisible to the human eye but detectable by AI.

The cloaking process does not visibly alter the image but makes it unusable for AI training purposes.

Attempts to counteract Glaze by programmers and prompters have been unsuccessful so far.

Glazed artworks appear unchanged to the naked eye, with no discernible difference between the original and the Glazed version.

Glaze can protect against image-to-image attacks, which involve modifying existing images with AI.

Glaze is an open-source tool developed by students from the U.S. and China.

The Glaze tool offers different settings to adjust the magnitude of changes and protection level.

Higher settings on Glaze provide stronger protection but may introduce more visible changes to the art.

Glazing an artwork may slightly distort its appearance, affecting the colors and overall look.

The Glaze process can be time-consuming, with higher quality settings taking longer to render.

Artists may need to fine-tune Glaze settings based on each individual piece of artwork.

Glaze offers a balance between maintaining the integrity of the art and providing protection from AI scraping.

For artists concerned about their work being used in AI training, Glaze provides a high setting for stronger protection.

The use of Glaze may require artists to inform their audience about the protection measures taken.

Glaze demonstrates the ongoing challenge of protecting creative works in the age of AI and machine learning.