After removing images generated by artificial intelligence models from its archives last month, Shutterstock is now going to sell ones created exclusively by OpenAI’s DALL-E 2 AI
dennizn/Shutterstock
Photo licensing service Shutterstock will begin selling images generated by artificial intelligence alongside those created by humans. The AI images will be powered exclusively by OpenAI’s DALL-E 2 software. Both companies say that human creators whose work inspired the AI will be compensated, but one artist describes the move as “sewer water leak into the drinking supply”.
Shutterstock was one of several photo agencies that began removing AI-generated art from their archives last month. A Shutterstock spokesperson told New Scientist that the company would continue to ban people generally from uploading AI-generated art to its platform, but that its collaboration with OpenAI was an attempt to embrace new technology in an ethical way. The two companies will also launch what the Shuttershock spokesperson called the Shutterstock Contributor Fund to “compensate artists for their contributions” and provide royalties when their intellectual property (IP) is used.
“When the work of many contributed to the creation of a single piece of AI-generated content, we want to ensure that the many are protected and compensated (instead of allowing an individual to generate and take full credit for that content),” they said.
Advertisement
Read more: Who should own the copyright on AI-generated artwork?
OpenAI didn’t respond to a request for comment, but the Shutterstock spokesperson said that a deal had been struck in which the AI that produces the pictures has been trained only on images from Shutterstock’s archives, rather than on content found elsewhere online.
Contributors whose work was used to train the models will receive a share of royalties from AI sales, said the Shutterstock spokesperson, but they didn’t say what percentage of revenue would go to contributors or how the contributions would be divided. It is often difficult to determine what input data was referenced to create any one piece of output.
Adrian Alexander Medina, editor of literary website and magazine Aphotic Realm and a creator of book covers, says he has lost three potential clients to AI-generated art since the start of October. He disagrees with Shutterstock selling AI-generated art and believes it risks ostracising photographers and illustrators.
“I work a lot with photo manipulation art. I pay fees for licensing and use completely royalty-free images. If websites like Shutterstock and others that offer licensed assets for graphic design use start allowing AI into their libraries, it’ll be akin to having sewer water leak into the drinking supply,” he says.
He is also sceptical of the company’s scheme to disperse revenue to artists from AI image sales. “I think it’s a feeble attempt to justify the stealing of millions if not billions of IPs that make up AI generated art,” he says.
More on these topics: