Last month, a group of visual artists sued Artificial Intelligence (AI) company Stability AI, Midjourney and DeviantArt for alleged copyright infringement.
This class-action lawsuit, filed by artists Sarah Andersen, Kelly McKernan and Karla Ortiz, is questioning the platforms’ use of Stable Diffusion, the AI image-generation system created by Stability AI, labelling it a “21st-century collage tool” that remixes the copyrighted work of millions of artists whose work was used as training data. According to a blog post on the lawsuit: “Stable Diffusion contains unauthorized copies of millions—and possibly billions—of copyrighted images. These copies were made without the knowledge or consent of the artists.”
Earlier this week, stock photo provider Getty Images sued Stability AI, accusing it of misusing more than 12 million Getty photos to train Stable Diffusion. According to a Reuters report, Seattle-based Getty has accused Stability of copying millions of its photos without a licence and using them to train Stable Diffusion to generate more accurate depictions based on user prompts.
Also read: Is NFT the disruptor that digital art needs?
These lawsuits, it seems, are only the beginning. The world of digital art is becoming complex as tools such as Midjourney and OpenAI’s DALL-E gain popularity. As more and more users try to give their artwork a touch of Vincent van Gogh, two big questions are staring at everyone: Whose art is it anyway and how will the law keep up with digital art?
To understand why these AI systems are coming under scrutiny, one needs to know how they work. These platforms train themselves on millions of digitised artwork to produce a new piece of art, based on text prompts, in seconds. Anyone can use them—from seasoned painters to amateur artists. Most offer a certain limit of free text prompts, followed by a subscription plan. London-based Stability AI, which also has a paid image generator service called DreamStudio, announced last October that it had raised over $100 million (around ₹820 crore) in funding and has been valued at $1 billion, a recent Reuters report noted.
This popularity is testing the limits of laws and ethics. “Legal issues relate to who is able to exploit the commercial rights deriving from unaltered images whose moral rights reside in the creator and copyright and other rights reside in either the author or the commissioner, depending on the jurisdiction,” says Noor Kadhim, a solicitor-advocate and art lawyer with over 14 years’ dispute resolution and arbitration experience. “If images are merely copied and, unlike in the educational or reportage exceptions, used to generate profit, then of course this will lead to lawsuits being started for damages/compensation. It is arguable they are already treading a fine line between what can be generated and what cannot,” Kadhim adds on email.
Also read: A virtual art experience
Copyright laws in the US and European Union don’t explicitly cover AI-generated art. Under Indian law, too, there’s no clarity. In a 2019 write-up, “Intellectual Property Rights Over Computer Generated Works”, on the firm’s website, Nandan Pendsey, of the law firm AZB & Partners, explains that “under Indian copyright law, typically the first owner of a copyright in a work is the ‘author’ of such work, subject to certain statutory exceptions. Further, such an ‘author’ in relation to any literary, dramatic, musical or artistic work which is computer-generated is ‘the person who causes the work to be created’. This provision was introduced in 1995, when the understanding of computer generated works was limited to those requiring human input. It is also pertinent to note that the term ‘computer generated work’ is not defined under the Copyright Act 1957.”
As Kadhim explains, the first question is: How do we even know which country’s laws to apply if this kind of art is decentralised and borderless? Further, when it comes to ethical implications, what if an artist doesn’t want certain images to be “diffused” in certain jurisdictions and the AI machine does so anyway? For example, images containing nudity or representations of religious figures. “The question becomes: Who is ethically exposed: the AI machine or the artist (who may be unaware)? Does it not then become a question of foreseeability—and what the artist could have foreseen by way of distribution when creating those works?” she adds.
It gets even more difficult. An AI-generated image can have multiple variations (or upscaled versions) and one user can even make changes to an image generated by another user through additional prompts. A recent Associated Press report explains how it can also be difficult to trace an image back to a specific tool or data source. “Some systems produce photorealistic images that can be impossible to trace, making it difficult to tell the difference between what’s real and what’s AI. And while some have safeguards in place to block offensive or harmful content, experts fear it’s only a matter of time until people utilize these tools to spread disinformation and further erode public trust,” the report added.
Artists are divided. Multidisciplinary artist Raghava K.K. has experimented with AI; his latest solo exhibition, The Impossible Bouquet, is based on similar lines and engages with prompt engineering to ask if we can shape our collective futures using this disruption as a springboard. “NFTs (non-fungible tokens) are threatening the existence of IP. Intellectual property is being questioned as a concept. There are these very fundamental shifts happening,” Raghava says on the phone. Referring to AI systems as an extension of the human consciousness, he says “we need to accept that we cannot be separated from our tools any more. We have to now engage the artists, the philosophers, the scientists to create experiments with reality…. I don’t see artists being replaced with technology. I see it as art being augmented by AI.”
Mumbai-based independent artists Samir Bharadwaj and Vishal Bharadwaj say some sort of middle path or stop-gap legalese will be needed. “I don’t think there has ever been a perfect scenario for artists and I don’t think there will continue to be. You can strengthen the laws to be more favouring towards human art. That’s the best you can do,” says Vishal. “There will be people and countries that will go to extremes. I am guessing at some point the French will have some kind of law completely protecting human artists,” he adds.
What can artists and digital art creators do in the meantime? Kadhim says they should be wary of not owning all parts of the work they are “creating”. “This goes down to the minutiae of the hardware and software used to create the art, licences granted and their scope, and to whom the images belong, including ‘images of images’,” she adds. “For example, you make a film and the film contains stills with a fridge in the picture, and on that fridge is a magnet with a work by David Hockney on it. You are liable as the digital art creator, for infringing copyright in that derivative work: right down to the reproduction of the magnet on the fridge in your film. So my advice would be: Make sure that everything that moves, makes a sound and contains a visual—you know the source of that and who owns it, and whether you need a licence for it.”