Photo: sawitreenan88/123RF
With the rise of AI, there’s been an increasing need for artists to protect their work from the datasets used to train the technology. And with the leak of a list of 16,000 artists used to train Midjourney, there’s been an outcry by artists for ways to safeguard their art. Until recently, the most popular method has been to “poison” images so they ruin the dataset. But now, thanks to Kin.art, creatives can ensure that their work never enters AI training sets.
Kin.art is an innovative portfolio platform that provides protection for artists’ entire inventory of work. So how does it work?
“As AI datasets require pairs of images and labels to be used in conjunction to properly train the AI on what the images are, if we can disrupt either the image or the label associated with a given artwork, we can prevent the artwork from being successfully inserted into the dataset,” explains Flor Ronsmans De Vry, CTO of Kin.art, who spearheaded the development of the platform.
“We disrupt both of these inputs, using image segmentation to ensure a complete piece of artwork is never inserted into an AI dataset, and using tag randomization to ensure that appropriate labels are never associated with each image,” Ronsmans De Vry continues. “This dual approach guarantees that artists who showcase their portfolios on Kin.art are fully shielded from unauthorized AI training of their work.”
One additional benefit of this lightweight solution to AI protection is that, unlike “poison” programs like Glaze and Nightshade, it doesn’t require any expensive hardware. Any artist who creates a portfolio on Kin.art will see their work protected efficiently and without an additional charge. Interestingly, Kin.art is also experimenting with using their technology to safeguard audio and video in addition to images, making it a comprehensive solution for creatives in all mediums.
While Kin.art’s technology is currently only available on the platform, Ronsmans De Vry says the company intends to release it to as many people as possible. Opening it up to third-party providers resonates with their belief that artists deserve to choose whether their art is included in AI datasets.
“Artists should not have to change their workflow to safeguard their work from AI; that burden should fall on the platform,” Ronsmans De Vry tells My Modern Met. “This is why we’re super passionate about the developer-first approach we’re aiming for, eventually allowing anyone to protect their own website against AI. We envisioned Kin.art as a safe space for creators to manage and grow their online presence. After we learned about the threat of AI, it was an absolute no-brainer for us to integrate the solution we came up with into this platform.”
Kin.art: Website
Related Articles:
Artist Recreates His Own Work With an AI Art Generator
AI “Completes” Keith Haring’s Intentionally Unfinished Last Artwork, Sparks Controversy
Getty Images Releases Commercially Safe AI Image Generator Based on Its Own Media Library
Celebs Warn About AI-Powered Deepfake Videos of Them Advertising Products on Social Media