Glaze Project

Roxane Lapa
11 Apr 202311:55

TLDRRoxy, the host, reviews 'Glaze', a tool designed to protect artists from AI style mimicry. She discusses the ethical issues surrounding AI-generated art, which often uses artists' styles without permission. Glaze adds a subtle distortion to artwork, making it less recognizable to AI and thus harder for AI to replicate the style. The tool is in beta version 3 and allows users to adjust the intensity of the distortion and render quality. Roxy tests Glaze on various pieces of her artwork, noting that the distortion is less noticeable on some pieces than others. She acknowledges the limitations of the tool, such as its inability to protect past works and the potential for AI developers to find a workaround. Despite these issues, she appreciates the effort behind Glaze and believes it could be useful for artists with a distinct style.


  • ๐ŸŽจ Glaze is a tool designed to protect artists from AI mimicry by adding a distortion layer to artworks, making it harder for AI to replicate the original style.
  • ๐Ÿ–ผ๏ธ The issue stems from AI tools using artworks from living artists without permission for training purposes, which Roxy previously discussed in detail in another video.
  • ๐Ÿ‘ฉโ€๐ŸŽจ Artists like Lois, Alana andami, Jacob Rosalski, and Greg Rutkowski have had their artworks unethically used in AI databases like Leon 5B.
  • ๐Ÿ› ๏ธ Glaze is currently in its beta version (version 3), focusing on improving AI resistance while minimizing visible distortion to human viewers.
  • โš™๏ธ The user interface of Glaze is straightforward but still has minor issues like overlapping elements and icon cutoffs.
  • ๐Ÿ” Users can adjust the intensity of the distortion applied to their artwork, with higher settings providing stronger AI resistance but more noticeable distortion.
  • โฑ๏ธ The effectiveness and processing time of Glaze depend on the user's computer power; better hardware results in faster processing.
  • ๐ŸŽž๏ธ The application supports batch processing, allowing users to apply the glaze effect to multiple images at once.
  • ๐Ÿ–ฅ๏ธ Glaze is available for both Windows and Mac platforms, ensuring broad accessibility.
  • ๐Ÿšง The effectiveness of Glaze may vary based on artwork type, with certain styles showing stronger distortions than others.
  • ๐Ÿ“œ There are ongoing legal and ethical debates about the use of artists' work in AI training, highlighting a potential need for future legislation or industry regulation.

Q & A

  • What is the main purpose of the Glaze tool?

    -The main purpose of the Glaze tool is to protect artists from style mimicry by AI. It does this by adding a layer of distortion to artworks, which aims to obfuscate the style when AI attempts to mimic it, resulting in generative art that does not replicate the original artist's style.

  • What prompted the development of Glaze?

    -Glaze was developed in response to concerns over AI tools scraping artists' works without permission to train their algorithms, which allows users to generate art mimicking the styles of living artists. This has raised ethical issues about intellectual property and the rights of artists.

  • How does Glaze work?

    -Glaze works by applying a distortion layer over an artwork before it is shared or published. This distortion is intended to be subtle enough not to be disruptive to the human eye but significant enough to mislead AI tools from properly mimicking the artist's style.

  • Can Glaze protect all types of artwork equally?

    -The effectiveness of Glaze varies depending on the type of artwork. Some artworks, especially those with complex or detailed styles, may show stronger distortions, while simpler artworks might not show as much difference.

  • What are the limitations of using Glaze mentioned in the transcript?

    -The limitations include visual distortions that may be off-putting to some viewers, and the fact that past work already shared or scraped cannot be protected by Glaze. Additionally, there's a risk that AI developers might find ways to circumvent the cloaking effect in the future.

  • What legal and ethical concerns are associated with AI generating art?

    -The legal and ethical concerns include the unauthorized use of living artists' styles and works to train AI, which can lead to financial and creative losses for artists. There are hopes for future legislation to address these issues after ongoing class action lawsuits.

  • What does the transcript say about the future development of Glaze?

    -The transcript suggests that while Glaze is promising and still in its early stages (beta version 3), its future effectiveness may depend on legal changes and technical advancements that prevent AI from bypassing the distortions applied by Glaze.

  • How do artists feel about AI tools like Adobe Firefly according to the transcript?

    -The transcript is critical of AI tools like Adobe Firefly, stating that despite claims of ethical development, these tools still use artists' works without providing opt-out options, which continues to raise ethical concerns.

  • What does the transcript suggest about the impact of Glaze on different art styles?

    -It suggests that the impact of Glaze might vary based on the art style. For example, the distortion on a black and white ink drawing was nearly imperceptible, whereas it was quite noticeable and less aesthetically pleasing on other types of artworks.

  • What are the personal feelings of the speaker, Roxy, towards using Glaze?

    -Roxy expresses that the visual distortion caused by Glaze, especially in its current beta version, is off-putting and would not be something she would use. She appreciates the intent behind Glaze but is cautious about its practical application in protecting artists' work.



๐Ÿ–Œ๏ธ Glaze: AI's Ethical Dilemma and Artist Protection

Roxy discusses Glaze, a tool designed to protect artists from AI-driven style mimicry, detailing the ethical concerns over AI using the styles of living artists without permission. She mentions notable artists like Lois and Greg Rutkowski, whose styles have been replicated by AI using data from models like Leon 5B. Glaze operates by adding a layer of distortion to artwork, which confuses AI without significantly altering the appearance to human eyes. This distortion prevents AI from accurately mimicking the artist's style. Roxy provides an overview of Glaze's functionality and tests the beta version's capabilities, highlighting both user interface issues and the potential of the tool.


๐ŸŽจ Testing Glaze on Diverse Artworks

Roxy experiments with Glaze on various pieces of her artwork to demonstrate how the tool's distortion affects different art styles. She notes that the distortion resembles overlays by abstract artists and varies significantly depending on the setting. Lower settings show minimal distortion, while higher settings result in drastic changes that could deter use. Despite testing with multiple artworks, she expresses skepticism about Glaze's utility in its current form due to the noticeable distortion. She also discusses the limitations of protecting previously shared or scraped artwork and the need for future legal and algorithmic changes to protect artists.


๐Ÿ”ฎ Future Challenges and Community Gratitude

Roxy addresses potential future challenges with Glaze, including the possibility of AI developers finding ways to bypass the distortions. She appreciates the efforts of the Glaze team but remains cautious about the tool's long-term viability. Despite the current limitations and the experimental nature of Glaze, Roxy sees potential benefits for artists known for specific styles. She concludes with gratitude towards the Glaze team for their dedication to protecting artists and thanks her supporters and viewers for their engagement with her content.




Glaze is described as a tool that aims to protect artists from style mimicry by AI. It works by adding a layer of distortion to artwork, which is not easily detectable by the human eye but confuses AI algorithms, preventing them from replicating the artist's unique style accurately. The script discusses Glaze as a defense mechanism against unethical practices where AI systems scrape artists' work without permission to train their models.

๐Ÿ’กStyle mimicry

Style mimicry refers to the ability of AI to replicate the unique artistic styles of living artists. This is a central issue in the video, as it highlights the unethical aspect of AI tools using artists' styles without compensation or permission. The script portrays style mimicry as a significant problem that tools like Glaze are trying to address, by making it harder for AI to accurately mimic these styles.

๐Ÿ’กGenerative art

Generative art is created with the use of autonomous systems, such as AI algorithms, which can synthesize art by learning from vast datasets of existing artworks. The video's script criticizes this practice when it involves using the works of contemporary artists without their consent, as it undermines their creative rights and financial livelihood.

๐Ÿ’กEthical concerns

Ethical concerns in the context of the video pertain to the morality and legality of using artists' work to train AI without permission. The video discusses how this practice potentially violates artists' rights and advocates for future legal reforms and class action lawsuits that could provide clearer guidelines and protections for artists against such exploitations.


Distortion in the video refers to the method used by Glaze to protect artwork. By applying a visual distortion that is subtle to the human eye but confusing to AI, Glaze aims to prevent AI from capturing the true essence of the artist's style, thus safeguarding it from being replicated and used without authorization.

๐Ÿ’กAI algorithms

AI algorithms are computational formulas that allow machines to perform tasks that typically require human intelligence. In the video, these algorithms are capable of creating art by learning from databases filled with scraped artworks. The script questions the ethical implications of such practices, especially when done without the consent of the original artists.

๐Ÿ’กClass action lawsuits

Mentioned as a potential future corrective, class action lawsuits in the video refer to legal actions taken by a group of artists against companies that use their work without permission to train AI. These lawsuits could lead to the development of laws that explicitly protect artists' rights in the digital age.


The video calls for legislation as a necessary step to formalize protections against the unauthorized use of artists' styles and artworks by AI companies. Such laws would aim to ensure fair compensation and control for artists over how their creative output is used.

๐Ÿ’กProtection against AI

This refers to the measures artists can take to safeguard their work from being exploited by AI technologies. The script showcases Glaze as a tool that adds protective distortions to artworks, complicating AI's ability to replicate styles accurately, thus acting as a barrier against misuse.

๐Ÿ’กArtistic rights

Artistic rights concern the legal rights and claims artists have over their creations. The video emphasizes the importance of respecting these rights, particularly in the face of new technologies like AI, which can easily and often unethically utilize artists' work without proper authorization or compensation.


Introduction to Glaze, a tool designed to protect artists from AI style mimicry.

Background on the controversy involving AI's use of artists' styles without permission.

Explanation of how Glaze works to distort artwork subtly to mislead AI training models.

Personal testing experience with Glaze Beta Version 3.

Discussion of UI issues encountered in the new beta version.

Detailed steps on how to use Glaze including setting distortion intensity and rendering quality.

Experimental results showing how different artworks respond to various distortion settings.

Observations on the varying effectiveness of Glaze depending on the type of artwork.

Mention of the potential for future AI technologies to circumvent Glaze's protection methods.

Ethical concerns highlighted about AI training practices.

Insights into Adobe Firefly's questionable ethics despite marketing claims.

The current limitations of Glaze, including its inability to protect past work already scraped.

Speculation about future legal and algorithmic changes that could enhance artists' protections.

Consideration of Glaze's potential value for artists known for a specific style.

Appreciation for the developers of Glaze and their intent to support the artist community.