An AI work from Ayaz Basrai’s ‘Solar Punk Cities’ series
When AI-generated ‘avatars’ from Lensa AI populated our feeds in the last couple of weeks, many were pleasantly surprised at how flattering the app had made them look, enhancing their best side profiles, with low-angle shots set to the backdrops of spaceships and gardens. Then came the truth, shared and re-shared by dozens of artists across the world: that some of these pictures were near carbon copies of copyrighted artworks of living artists. In some, there were even remnants of mangled signatures of the original artist, so blatant was the art theft.
What happens when art has quite literally been scraped off by AI without crediting the artist or getting their consent? According to Ayaz Basrai, an industrial designer and architect based in Goa, the notions of authorship, ownership, plagiarism and inspiration already have very blurry lines and were contested ideas much before the advent of AI. “I believe the output of text-to-image systems has called this sharply into focus,” says Basrai, who has been experimenting on MidJourney. Some of his popular works include ‘Solar Punk Cities’ where he has depicted visuals of Indian cities through a dystopian lens. “The structure of the Internet is redundancy of copies of copies of copies. A fungible [non-encrypted] piece of digital art is probably one of the easiest to copy, replicate or scrub into a database.”
Policing the use of digital art, he adds, is an extremely resource-intensive and fairly opaque process with limited success. And the companies responsible for the theft definitely know this is the case.
In the Indian context, our copyright law is structured along the lines of the Berne convention (the protection of works and the rights of artists). “According to it, if you are generating derivative works from an artist, you need to take permission from the artist,” says Sandhya Surendran, a media and entertainment lawyer based in Bengaluru. “In the case of Lensa AI, they have taken artworks from different artists and fed them into the AI model. This becomes derivative work and they are supposed to take permission. And not just in India, but also across Europe and the United States because they are all signatories to the Berne convention.” The only exception to the rule is fair use, where artworks can be replicated for entertainment or news purposes.
While artists do have the right to legal remedy, the actual enforcement is tricky. Unlike the West, where legal recourse is generally swift and prompt, that’s not the case in this country. Artists will hesitate to file cases because of the operational constraints that come with it in India — the waiting, the navigating a clogged legal system, and the waste of time and resources. “Artists often don’t have the bandwidth to go through all of this,” she adds.
There are a rare few companies navigating AI in an ethical way. “We have Bengaluru-based Beethoven that pays artists so that their works can be fed into their AI model,” says Surendran. “This workflow needs to be followed.”
A reductive lens
Basrai believes that the storytelling around speculative fiction is the main event, not just the image. “As a disclaimer, I must state that I am not a digital artist. For me personally, as a dabbler in Stable Diffusion and Midjourney, it’s an opportunity to interact with another intelligence — a symbiotic collaboration, and not necessarily as a threat,” he says. “I’d rather see this [AI-based] form of plagiarism as the most recent step in a very old series of plagiarism. Ettore Sottsass’ Memphis movement was copied from houses in Tiruvannamalai, and Picasso lifted ideas around Cubism from African indigenous masks.” He believes that if we can collectively view culture as progressive or regressive, this may offer us a richer lens. “The who-copied-from-whom lens is sort of reductive and I don’t think it defines the human experience in any meaningful way.”
Jon Lam, a Canada-based digital artist, moodboard storyteller and animator — who has worked with various clients including Marvel’s Miss Marvel — says artists must engage with their local politicians to seek remedies and find ways to protect their art from unethical usage by AI. Lam was amongst the first group of artists to publicly call Lensa out, and his opinions were later featured in The New York Times and washington postamong others.
“People say there are some quick fixes such as typing robot.txt in your website domain to protect yourself, but these are band-aid solutions,” he says. “Now that the media has made us all aware of how AI will affect not just artists but also lawyers and politicians, there is a need to engage with all stakeholders before major companies start lobbying for their products.”
On an individual level, how does an artist who just shares their artwork on Instagram protect themselves? Lam suggests putting “loud and obnoxious” watermarks on critical areas of the artwork (like hands or key elements). “You can always share a lower resolution image of your artwork on social media and link your viewers to your password-protected website, though some scraping will even bypass your password. You need to make your voice heard on every platform possible.”
The writer is an author and editor based in Mumbai.