Shutterstock Expands Partnership with OpenAI to Build Next-Generation Generative AI Tech
The happenings in the AI visual industry keep coming! Shutterstock announced today that they expanded their collaboration with OpenAI –the lab tech behind AI image generator Dall-E 2 and AI chatbot ChatGPT– for several years into the future.
The long-term partnership lets Shutterstock have priority access to OpenAI’s AI models to expand and improve AI features on their service. OpenAI, on its part, will get to use Shutterstock’s extensive media catalog and metadata to enrich its AI training datasets. Plus, they’ll leverage one of Shutterstock’s latest acquisitions to bring AI image generation to mobile platforms.
Find out all about this next!
Shutterstock and OpenAI's New Multiyear Agreement
Shutterstock and OpenAI initially partnered back in 2021, and the first fruits of their collaboration came through at the end of 2022 when the stock photo agency unveiled Shutterstock Generate, their native AI image generator built on the Dall-E model but trained solely with Shutterstock content.
The newly signed deal sees both companies collaborating for another six years, with a clear strategic plan to further generative AI capabilities: bringing more and better AI functionality to Shutterstock's website and giving OpenAI better and improved training datasets.
Shutterstock Leverages Dall-E to Expand AI Image Editing
As part of this agreement, Shutterstock will maintain preferential access to OpenAI’s latest Dall-E model and enrich their generative AI offer for users –which right now consists of an AI image generator that is legally safer due to being trained with approved, licensed content only– with new AI image editing functionality.
This will allow Shutterstock customers to synthesize entirely new images as well as transform any image in the agency’s library to meet their creative needs.
Plus, they will be introducing Shutterstock Generate through their API service, letting customers integrate these capabilities into their workflows easily.
OpenAI to Train their Models with Licensed Shutterstock Content
At the same time, this partnership lets OpenAI secure licenses to Shutterstock’s massive media catalog, which is high-quality, metadata-rich, and cleared for AI training usage. Furthermore, through the Contributor Fund they have in place, all Shutterstock contributors receive monetary compensation in the form of royalties for the use of their work in AI training and AI image generation.
This way, the tech lab can now train its models with ethically-sourced datasets that are fairly compensating the human artists behind them.
This is a pretty relevant fact given how Dall-E competitors Stable Diffusion and Midjourney are being sued over using web-crawled images in their datasets – and Dall-E was presumably doing similar until now.
Companies Coming Together to Bring AI Image Generating to the Mobile Sphere
Finally, the announcement expressed that Shutterstock and OpenAI plan to collaborate to exploit GIPHY –Shutterstock’s latest acquisition–, particularly its extensive mobile platform and user base, to develop generative AI features accessible from mobile devices.
This next-generation step will give people new ways to express themselves through visual media.
It’ll be pretty interesting to see how those new features in Shutterstock come along and what changes, if any, happen on Dall-E through new training strategies. And what exactly will this new, GIPHY-powered mobile AI model be? Time will tell!
