Getty’s AI-generated image tool allays some artist concerns
The image generator helps the stock image company get into the generative market and answer copyright questions for businesses but could still be a concern to artists.
Getty Images' new AI-powered image generator eliminates some problems with AI-generated images but fails to address others.
Getty on Sept. 25 revealed that it has partnered with AI hardware and software vendor Nvidia to launch Generative AI by Getty Images. The AI-powered image generator is trained on Edify models in Nvidia Picasso and uses only images from Getty's library. Getty customers creating and downloading visuals with the tool will receive a royalty-free license.
Getty also assured customers that images created with the tool will not be added to the Getty or iStock libraries. Contributors will also be compensated for any work on training the tool.
Getty's lesson
The new AI image tool comes seven months after Getty sued Stability AI for allegedly using the images in Getty's system to train Stability AI's Stable Diffusion tool.
"It looks like they have taken Stability to task," said Michael Bennett, director of education curriculum and business lead for responsible AI at Northeastern University. "They've learned not only the lesson that's evident in that lawsuit -- that they need to check the set of images the AI companies use to train their AI image generators -- but [also] … that they could get into the same game."
This is Getty's way of protecting its market while taking advantage of new generative AI developments, Forrester Research analyst William McKeon-White added.
"It is a good way for Getty to reassert IP ownership and leverage their existing libraries," he said.
The benefit for artists and businesses
For artists and businesses who don't yet feel comfortable with the current state of AI-generated images, this may be the best they can expect, Bennet added.
They've learned not only the lesson that's evident in that lawsuit -- that they need to check the set of images the AI companies use to train their AI image generators -- but [also] … that they could get into the same game.
Michael BennettDirector of education curriculum, business lead for responsible AI, Northeastern University
"There will always be folks who feel that any kind of commercialization of your work, especially if it's a corporatization strategy, is unfounded and unfair and indefensible," he said. "For those who were comfortable having Getty act as their representative and license their work to commercial fashion … this new system ought to make them feel much better."
For some businesses, Bennett said, this removes potential worries about using copyrighted material and makes it less costly should they decide to eliminate some artists within their company.
For Getty, it needs a partnership with Nvidia because of the computing power the AI vendor can provide and because there's no conflict of interest between the two companies.
"A partnership with Nvidia, as many are doing, is smart -- to work with a company on the leading edge where financial incentives are also aligned," McKeon-White said. "Nvidia isn't interested in encroaching into Getty's market and vice versa. I couldn't say the same for all other providers."
The other ethical question
Despite this benefit for Nvidia, Getty, and even some artists and businesses, Bennett said an ethical question still exists.
"It does not remove the potentially larger threats of generative AI systems encroaching on what had been the exclusive workspace of human artists," he said.
While this move from Getty seemingly eliminates some copyright problems and other problems that could arise for businesses, for artists, this is just a new way of treading on their profession.
"There's still the ethics problem and the labor problem that you're contributing to -- building this AI, this generative AI system that's going to replace artists in the short- and the long-term," Bennett said.
Another challenge for Getty is meeting the various needs of users using this tool, McKeon-White said. Those needs include having different ways of changing image outputs as well as ensuring there's no bias in the system and that Getty continues to update its data.
Esther Ajao is a TechTarget Editorial news writer covering artificial intelligence software and systems.
