Shutterstock announced this week that it has partnered with OpenAI to introduce a AI text-to-image solution for its customers. This comes just a month after Getty Images decided to ban AI-generated art altogether.
Sailing in Different Directions
Getty Images may not think of Shutterstock as a direct rival. Shutterstock has more than 450 million images on its site and carries everything from low-end clip art to high-end photography. By contrast, Getty Images has 80 million images and tends to focus on higher-quality images while offering both royalty-free and novel licensing models. I’m sure they would say it is a more curated experience.
Getty’s ban rationale was based on what it characterized as unexamined rights issues. “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,” says Getty.
Shutterstock decided to take a more constructive viewpoint. “The mediums to express creativity are constantly evolving and expanding. We recognize that it is our great responsibility to embrace this evolution and to ensure that the generative technology that drives innovation is grounded in ethical practices,” said Paul Hennessy, Chief Executive Officer at Shutterstock.
The company says it will “launch its AI-generated content capabilities in a manner that is responsible and transparent for its customers and contributors. In expanding its partnership with OpenAI, launching a fund to compensate artists for their contributions, and focusing its R&D machine on gathering and publishing insights related to AI-generated content, Shutterstock positions itself at the forefront of emerging technology -- supercharging ethical, creative storytelling for all.”
A Constructive Approach
You have two stock image websites with a completely different approach to adopting new AI technology. Getty plays it safe and seeks to protect its high-priced and exclusionary business model. Shutterstock likely sees the rise of text-to-image technology adoption as both inevitable and a threat to its business and decides to take a constructive approach.
How Shutterstock plans to compensate artists for their contributions is unclear. And the company does not comment on whether it is talking just about images created by artists using text-to-image solutions or images generated by others that have a resemblance to art created by a specific artist. It seems like the former, but there may be room for the latter if there is ever a way to assign provenance in part to a previously created work.
Easy Access
Shutterstock says it will make DALL-E 2 image generation available through its website in the coming months after integration is complete. This is a good strategy for OpenAI. Shutterstock has far more reach than OpenAI and specializes in making solutions available to novice, less adventurous users. Shutterstock will eliminate the need for OpenAI to educate prospective users while significantly raising awareness.
It is also becoming clear that there is a skill to using text-to-image generators well. Anyone can create an image with generative AI software, but not just anyone can create something that is very good. So far, a few YouTubers and bloggers have been offering the market some education about how to get more out of the models. Shutterstock may be able to do something similar on a larger scale.
The news is also a win for OpenAI as it is facing stiff competition, especially from Midjourney and the open source models of Stable Diffusion. There is no dominant player yet in the text-to-image generation space, and that means larger platforms that could drive a lot of use and user education can significantly influence vendor market share.