Shutterstock has unveiled plans to expand its partnership with OpenAI, providing the startup with training data for its AI models while positioning its own library as a place where generative AI can be used inside a licensed stock-content business.
Under the agreement, Shutterstock and OpenAI will partner over the next six years. OpenAI will obtain images, videos, music, and associated metadata from Shutterstock. In return, Shutterstock customers will receive “priority access” to OpenAI’s latest technology and new editing capabilities intended to transform images in Shutterstock’s stock content library.
That exchange captures the central bargain now forming around generative AI. AI companies need large, varied bodies of material to train and improve their models. Content platforms, meanwhile, face a threat from tools that can produce an image on demand instead of sending a customer to search through a conventional library. A licensing arrangement does not remove that tension, but it gives Shutterstock a way to participate in the shift rather than simply watch it erode the market it helped build.
OpenAI and Shutterstock have renewed and significantly expanded their strategic partnership. The relationship is not confined to still images, either. Through Giphy, the GIF library Shutterstock recently acquired from Meta, OpenAI will bring its generative AI capabilities to mobile users.
The Giphy element matters because it moves the partnership beyond the familiar workflow of a designer searching for stock photography. Mobile users encounter visual media in faster, more casual settings: messaging, social posts, short-form communication, and everyday image editing. Generative AI in that environment is less likely to feel like a specialist production tool and more like a built-in creative option. For Shutterstock, that creates a potential distribution path far removed from the traditional stock-library search page.
“This partnership reinforces our commitment to driving AI tech innovation and positions us as the data and distribution partner of choice for industry leaders in generative AI,” said Shutterstock CEO Paul Hennessy in a press release.
Hennessy’s wording is revealing. Shutterstock is not presenting itself only as a seller of individual images, clips, or tracks. It is also making a case that its collection, metadata, contributor relationships, and customer base are valuable infrastructure for companies building AI products. In a market where the provenance of training material has become a major concern, a formal deal can be commercially useful even before courts settle the larger legal questions.
Table of Contents
The uneasy economics of stock content and generative AI
Shutterstock and generative AI startups have a tense relationship. Generative AI, especially generative art AI, poses a serious challenge to stock galleries because of its power to instantly generate highly customized stock images.
For a buyer, that appeal is straightforward. Stock libraries have historically offered speed, breadth, and legal clarity, but they still require a search process and some compromise. A generated image can be tailored around a particular composition, subject, mood, or format. That does not mean it will always replace licensed stock content; customers may still need real-world footage, recognizable editorial material, specific visual assets, or confidence in the source of an image. But it changes the baseline expectation. The question is no longer simply whether a library has the right image. It is whether a user can create a close-enough alternative immediately.
That pressure is especially acute for the people whose work fills stock libraries. Generative AI startups have come under fire from contributors to stock image galleries, such as artists and photographers, for what they perceive as an attempt to profit off their work without giving credit or compensation. The dispute is not merely about technology. It is about who has the right to benefit when a system is trained on creative labor and can then generate material that competes with the people who made that labor possible.
Getty Images recently filed a lawsuit against Stability AI, the creators of the AI art tool Stable Diffusion, for scraping its content. Stability AI allegedly unlawfully copied and processed millions of Getty Images submissions protected by copyright to train its software. Consequently, a trio of artists is suing Stability AI and Midjourney, an AI art creation platform, for copyright infringement.
Those disputes have become a backdrop to nearly every major commercial agreement involving generative imagery. They also expose a divide between two approaches to sourcing data. One approach treats publicly accessible material as available for model development unless a court decides otherwise. The other seeks licenses, contracts, or platform-level arrangements before the material is used. Neither path ends the policy debate, but the second offers companies a clearer commercial story to tell customers and contributors.
However, some experts believe that training models using public images, even copyrighted ones, will be justified under the fair use doctrine in the U.S. That view remains important because it suggests that the legal status of training data may not map neatly onto ordinary copying disputes. At the same time, the fact that fair use arguments exist does not make the issue commercially settled. Companies still have to manage litigation risk, public criticism, contributor trust, and the practical need to explain where their AI-enabled products come from.
See Also: OpenAI launches GPT-4 model
A partnership built around access and control
It is unlikely that this issue will be resolved soon, so Shutterstock, seemingly not wanting to depend on a prolonged court dispute for profits, has partnered with OpenAI to launch an image creator that is powered by OpenAI’s DALL-E 2.
This partnership between Shutterstock and OpenAI dates back to 2021, and the image creator was launched in late 2022. The expanded agreement therefore looks less like a sudden response to the latest AI frenzy than a deeper commitment to a direction Shutterstock had already chosen. Its strategy is not to argue that generative tools can be kept outside the stock-content business. It is to put those tools within a platform that can govern access to source material, offer editing features to customers, and create a compensation mechanism for contributors.
That distinction could matter for buyers as much as for Shutterstock. Businesses adopting generative tools often need more than a striking image: they need a workflow, a known supplier, and terms they can understand. “Priority access” to OpenAI’s latest technology may help Shutterstock make its service feel less like a passive archive and more like a working creative environment. New editing capabilities could also preserve the usefulness of existing stock material, allowing customers to adapt assets rather than discard them when they need a variation.
OpenAI, alongside Shutterstock, has formed licensing agreements with Nvidia, Meta, LG, and other companies in order to construct generative AI models and tools across 3D models, images, and text. The breadth of those arrangements points to a larger reality: generative AI is not one product category. It touches visual creation, language, media libraries, and the systems used to distribute and modify digital assets. Partnerships can give AI companies data and reach, while giving established technology and media companies a role in the value chain.
The contributor fund is the hardest part of the promise
In addition, to express appreciation for the artists who work on its platform, Shutterstock has created a “contributor fund” that compensates artists for their efforts in training Shutterstock’s generative AI and provides them with ongoing royalties for future licenses of newly-generated assets.
The fund is significant because it acknowledges the core objection from contributors: participation cannot be treated as an afterthought once a model has already been built. Compensation and ongoing royalties do not end the arguments around consent, credit, or competition. They do, however, set a different expectation from a model in which creators see their work used to train commercial systems without a visible route back to payment.
Shutterstock’s bet is that licensed data, access to OpenAI technology, and contributor payments can coexist in a business increasingly shaped by generative creation. It is an ambitious position, and it will be tested by the same forces that made the partnership necessary: customer demand for faster visual production, artists’ demands for fair treatment, and unresolved legal questions over copyrighted training data. For now, the company has chosen engagement over resistance—and has tied that choice directly to OpenAI’s expanding AI ecosystem.
