It is evident that artists are responding favourably to Nightshade’s campaign against AI scams.
On January 18, following several months of research and peer review, Nightshade, a programme that stops artificial intelligence art generators from plagiarising content that is visible by the public, became available to the public. The fact that painters, photographers, and other individuals who share photographs online are keen to protect their work from generative artificial intelligence has become abundantly obvious since that time. Nightshade has been downloaded a quarter of a million times in the five days that have passed since its initial release, according to the organisation that is responsible for the game.
“I expected…extremely high enthusiasm,” said Ben Zhao, the manager of the Nightshade project and a professor of computer science at the University of Chicago, in an interview with VentureBeat. “However, I continued to underestimate it…The reaction is simply beyond everything that we could have hoped for. The downloads are coming from all around the world, according to the responses on social media.
Even if you have copyrighted your work, any artwork or photographs that you upload to the internet will be incorporated into a dataset that is used for the training of generative artificial intelligence. Running your photographs through a data-poisoning application is the only way to stop artificial intelligence art generators from stealing your material when they are used to create art. The Nightshade plant is a good example of this. After you have uploaded an image to Nightshade, the programme will modify the data that is built into that image, rendering it unsuitable for use in providing training for models. Only works of art that lack a significant amount of texture may disclose a few odd pixels here and there if the artist has omitted to use Nightshade’s “low intensity” setting. The majority of “shaded” images appear to the human eye to be the same as they did before Nightshade was introduced because of this.
When you shade an image, you inhibit the data scrapers that are used to feed AI models from correctly interpreting the contents of the image. On their website, the developers of Nightshade present an example of a cow in a green field, which may appear to artificial intelligence to be analogous to a leather purse that is laying in the grass. The image of a dog transitions into the image of a cat, and so on and so forth. Even if a shaded image has been cropped, compressed, or graphically manipulated with a programme such as Photoshop, generative artificial intelligence technologies such as DALL-E and Stable Diffusion are unable to determine what the original image contains.
For certain artists, Nightshade may not be sufficient to prevent the soiled hands of artificial intelligence from touching their work. This is a crucial point to keep in mind. It is necessary to use a different tool in order to avoid artificial intelligence from imitating the style of an artist, even though Nightshade modifies the content of an image behind the scenes. Glaze is Nightshade’s older sibling and the original tool developed by the University of Chicago team. Glaze operates in a manner that is comparable to that of Nightshade in that it has the ability to edit the data of an image without significantly altering its appearance. However, rather than giving the impression that a piece of artwork represents a different subject, it alters the visual style of the piece. To an artificial intelligence, a charcoal portrait, for instance, would appear to be a contemporary abstract Jackson Pollock imitation. Utilising both Nightshade and Glaze is therefore a more effective way to guarantee that an artist’s work will continue to be unusable by artificial intelligence data scrapers.
The Nightshade team has also stated that they are less concerned with destroying artificial intelligence art generators and more concerned with refocusing artists. According to the information that can be found on their website, “Nightshade’s goal is not to break models; rather, it is to increase the cost of training on unlicensed data, so that licencing images from their creators becomes a viable alternative.”
There is no cost associated with downloading Nightshade or Glaze from the website of the team. Additionally, they can be obtained through this X post, which was created by the developers after the numerous downloads of Nightshade caused their campus network link to become saturated. (It is important to remember that clicking on either of the URLs featured in this page will immediately begin the download process.) A tool that integrates the functions of Nightshade and Glaze is going to be developed by the team during the next few months. This tool will make both of these applications easier to use together.