With a new world of visual potential lately opened by artificial intelligence (AI) image generators such as DALL-E and Midjourney, there is also a new world of potential legal complications. Looking to get ahead of problems before they begin, Getty Images — a massive supplier of stock photography — has banned the upload and sale of all content generated using AI art tools. The technology can quickly generate multiple takes on imagery from user-supplied text prompts, with results that range from silly, to pretty realistic, to fairly nightmarish, to truly the worst thing ever.
But where is the source material for these AI bots coming from? For the most part, it’s being scraped and remixed from the work of human artists, who use the Internet as a venue for their work to connect with audiences and potential buyers. Not only do some see this as disenfranchising to artists who have worked hard to develop a personal brand, but it also presents legal quicksand for image sites that decide to trade in AI-crafted content.
“Effective immediately, Getty Images will cease to accept all submissions created using AI generative models (e.g., Stable Diffusion, Dall-E 2, MidJourney, etc.) and prior submissions utilizing such models will be removed,” reads a statement circulated this week by the company to media and its image providers. “There are open questions with respect to the copyright of outputs from these models and there are unaddressed rights issues with respect to the underlying imagery and metadata used to train these models.”
The statement went on to clarify that the limits on submissions do not prevent 3D renderings or impact the use of digital editing tools like Photoshop and Illustrator.
When asked by Hyperallergic how many images currently on the website will be impacted by the new policy, Craig Peters, CEO of Getty Images, said: “To our knowledge, it is extremely limited within our creative content library and there were already significant controls for our editorial offering. We are communicating with other companies and communities to understand perspectives with respect to these issues, how the legal or regulatory bodies might address and whether we might be helpful to resolve.”
“Our entire purpose is to bring creatives together into a safe, honest, and vibrant community to create fantastic images, and so the use of 100% machine-generated images, whilst an incredible breakthrough, is not something that helps our community,” reads a September 14 statement by PurplePort CEO Russ Freeman about banning AI art on the site. “I feel that using machine-generated images, while empowering everyone to participate in generating art, does not reflect the core purpose of our service, nor does it contain enough human input.”
Peters told the Verge that Getty Images will rely on users to identify and report such images, and that it’s working with C2PA (the Coalition for Content Provenance and Authenticity) to create filters screening for AI content. Other major image stock providers, like Shutterstock and Adobe Stock, have not yet explicitly banned AI content for sale, but the entire industry is attuned to how current discourse about artistic appropriation will play out across regulatory law, which often tends to trail tech innovation by years or even decades.
That leaves a lot of meantime for the digital art space to continue to be inundated with strange, proliferating many-eyed corpse babies, which seems to be the aesthetic space where AI art is always thriving — and a lot of wasted time for artists trying to protect their IP by playing a losing game of whack-a-mole against the internet.