Stay up-to-date with our daily and weekly newsletters, providing exclusive industry-leading AI content.
The rise of generative AI has led to the creation of numerous image generation models trained using data taken, often without permission, from artists’ works found all over the internet. Some artists are now actively looking for ways to protect their creations from such practices.
Enter Kin.art, a new tool designed to safeguard not just a single image, but an artist’s entire portfolio. Kin.art is part of a new online art hosting platform that offers rapid, built-in defenses against AI for any image an artist uploads to its servers.
Kin.art’s co-founder and CTO, Flor Ronsmans De Vry, announced today that their defensive method differs from existing solutions like the University of Chicago’s Glaze Project. While Glaze and its recent tool, Nightshade, aim to confuse AI models after artwork has been included in a dataset, Kin.art prevents the inclusion from happening in the first place.
Kin.art employs two machine learning techniques: image segmentation and label fuzzing. Image segmentation breaks apart an image and analyzes each piece, scrambling it to appear disorderly to machines while still looking normal to humans. Label fuzzing scrambles the metadata attached to images, disrupting AI training algorithms. This dual approach makes it difficult for AI to correctly understand and learn from these images.
Kin.art aims to be the first line of defense for artists, ensuring that their artworks are never accurately ingested by AI training models. The platform is free for artists to use; they simply create an account and upload their works, with the option to turn the AI protections on or off.
In terms of revenue, Kin.art plans to charge a small fee on artworks sold or monetized through their e-commerce features. This will help fund the platform’s growth and enable continuous support for the artist community.
Upon its public launch, Ronsmans De Vry provided further insight into Kin.art’s functionality. For example, while all images undergo the same segmentation and fuzzing process, the mutations are unique to each image. This process takes only a few milliseconds and doesn’t affect how artworks are viewed online, unless someone tries to download them without permission.
Artists can opt-out of these protections via a simple toggle during the upload process. Using Kin.art remains free, and the platform respects artists’ preferences, allowing them to control whether their works have AI protection.
Currently, Kin.art is focused on refining its product and has not yet built a substantial user base. Existing customers can opt to enable AI protections whenever they choose to.
As for the name, “Kin.art” combines meanings from both English and Japanese, symbolizing family and gold, respectively, reflecting the platform’s aim to create a valuable community of artists.
Kin.art also accommodates AI-generated artworks, recognizing the need for both human and AI art to coexist while ensuring clear labels to avoid misrepresentation.
Facing competition like Nightshade, Kin.art positions itself as a preventative measure, making it a valuable option for artists looking to protect their work before any scraping can occur. While Nightshade can take considerable time and resources to apply, Kin.art’s rapid approach offers a more scalable solution.
For the press coverage query, the mentioned media platforms have previously covered the co-founder team, thus being included on Kin.art’s homepage. Links to specific articles were provided by Ronsmans De Vry.