Under Siege

In 2022, generative AI was a novelty. By early 2025, billions of AI images were being generated per month. Adobe alone reports that its Firefly image generator created three billion images within months of launch, surpassing the total archives of many traditional photo libraries. Suddenly, what seemed like an unshakeable $5 billion stock photography market, dominated by Getty Images and Shutterstock, faces very real pressure.

At its peak, the stock-photo business felt settled; it was a five-billion-dollar pool of predictable royalties largely controlled by Getty Images and Shutterstock. The article looks at how the large stock-image companies reacted and responded to the surge of AI-generated images, including early lawsuits, a three-billion-dollar merger and cooperation with AI companies. Finally, the article asks what, if anything, still anchors the old stock-photo model. Who needs it, will it survive, and, if so, how?

Closing the Gap on Photorealism

Image generators have moved quickly from novelty to near-indistinguishable realism. Early 2023 outputs still gave themselves away with oddly shaped hands or rubber-looking textures. Twelve months later those tells were largely ironed out, and by the end of 2024 a Conjointly study found that most people scored little better than a coin-flip when asked to spot the real photograph.

This change matters because a large slice of stock-photo demand has always been surface-level: product placeholders, generic office scenes, conceptual collages. Buyers in that tier care about cost, speed, and rough fit, not provenance. If a prompt delivers a custom mock-up in ten seconds and viewers cannot tell the difference, the incentive to pay for a traditional licence fades fast.

The agencies know it. Contributors on microstock forums report falling download counts for the most ordinary subjects such as hands typing on laptops, smiling business teams, isolated objects on white. Those are precisely the images easiest to replace with a prompt.

Revenue Shifts and Licensing Deals

So how are these companies performing now that the bread-and-butter image purchases are disappearing? The first number is counterintuitive, but also slightly misleading; Shutterstock reported $935 million revenue in 2024, up 7% from the previous year. This would be somewhat remarkable, given the backdrop of AI growth, so it's not altogether surprising that over $100 million of this came from licensing image data to generative AI companies. Shutterstock's deal with OpenAI alone could generate as much as $250 million by 2027, a significant pivot away from their core business of selling photos. To help us see the impact of AI, Getty Images acknowledged that its flagship "Creative" stock segment declined by nearly 5% year-on-year in 2024 (again this was offset by gains in editorial and similar AI data licensing agreements).

From Litigation to Partnership

Once the capabilities of AI image generators became clear, the stock agencies naturally saw AI as a threat. Their first move was, of course, legal: haul the new image-makers into court and try to form a resistance through litigation. In early 2023, Getty sued Stability AI, accusing them of copyright infringement for scraping their content to train image generators. The case is still working its way through the courts, in fact as of writing, the early discovery work has barely finished. So as of yet nothing is settled, and we may not know more for some time. Winning, for Getty, would require proving that bulk scraping is not fair use, something US courts have never ruled on at this scale.

Various other lawsuits underscore how unsettled the legal terrain remains. Alongside Getty v Stability AI, illustrator Sarah Andersen and fellow artists have sued Stability AI, Midjourney, and DeviantArt for unlicensed image scraping; The New York Times has filed against OpenAI and Microsoft over the reuse of its archives; the Authors Guild leads a class action claiming mass ingestion of copyrighted books by OpenAI; and Universal Music, ABKCO, and Concord have taken Anthropic to court for using song lyrics in model training. Each case raises the same core question still awaiting a decisive ruling: does bulk scraping of copyrighted material for model training qualify as fair use? Until a judge answers that directly, every plaintiff, Getty included, is litigating in slow motion while the market moves on.

Getty's lawyers are accurately aware that such a decision, let alone a favourable judgment for them in court could arrive years after the market has shifted. And all the time, tech firms continue to train on every available data source, and their clients are experimenting with new tools. Faced with such a reality, Getty and Shutterstock chose a parallel track: aggressively monetising their image libraries by licensing them directly to the AI platforms they had previously opposed, rather than wait for an uncertain court victory. The lawsuit continues, but no longer as the centrepiece of their strategy; it is now something of an insurance rather than plan A. This is profitable for now, but it's unclear how long this revenue stream can last. AI companies initially needed vast libraries to train their image generators, but once trained, these models might not require continuous access to such vast amounts of fresh content. Shutterstock itself has acknowledged that recurring AI licensing revenues might plateau soon.

A Merger Born of Tightening Margins

Getty and Shutterstock have followed largely parallel tracks for years, each claiming roughly half a billion dollars in annual image-licence sales and maintaining catalogues that together exceed 700 million files. By late 2024 both were also carrying higher cloud-storage bills, rising contributor payout obligations, and a new reliance on one-off data deals with AI developers. Investors started asking how long the business could rely on quick fixes once regular image downloads stopped growing or began to fall.

Rather than fight for the same shrinking pie, the boards signed off on a full merger announced January 2025. The deal is valued at about 3.7 billion dollars, with a stated goal of stripping out 150-to-200 million in duplicate overhead during the first three years. Management says the larger catalogue will improve pricing power with enterprise clients and put them on stronger footing when negotiating bulk data licences with the major model builders. In short, the combined firm is betting that scale plus cost discipline will buy time while it figures out a new, post-licensing business model.

Who Still Buys Stock Photography?

Two years ago (pre-Gen-AI), most stock-photo downloads still came from a familiar spread of small agencies, freelance designers, and self-serve marketing teams. Generative tools changed that almost overnight. Surveys in early 2024 showed nearly forty percent of marketers were already using prompt-based systems for day-to-day social posts and banner ads. Fast, cheap, and infinitely customisable beats a search box full of "close enough" choices, so that entire long-tail segment has drifted away from paid licences.

Financially, though, those buyers were never the engine of Getty or Shutterstock. Their entry-level subscriptions carried thin margins and little pricing power. The real money sits with large companies that cannot afford visual mistakes. Banks, hospitals, and insurers need images that pass legal review without debate. Travel brands want real hotel rooms and real beaches, not synthetic placeholders that risk a mis-sell claim. News outlets require verified shots of actual events. Getty's editorial archive meets that need and the figures show it: editorial revenue rose while creative stock slipped.

So, yes, the low-value customers have more or less exited, but the high-stakes clients that require and pay for provenance, indemnity, and authenticity are still writing big cheques.

Evolving from Static Archives to On-Platform Editing

Both agencies have shifted from simply selling finished files to offering on-the-fly AI-driven alterations inside their portals. Shutterstock added a prompt-driven tool that lets a user retint a product, swap colours, or erase a background object in a single step. Getty released a similar feature for its enterprise clients, pitched as a way to keep brand teams and compliance officers in the same workflow. What matters is not the novelty of the edits, any competent designer could do them in Photoshop, but the legal wraparound: every change is logged, tied back to the original licence, and covered by the same indemnity that guarded the unedited photo.

This audit trail is the part CFOs and legal teams care about. A marketing manager can show that the final asset stayed inside a controlled environment from download to publication, avoiding the grey zone that comes from exporting a file, tweaking it offline, and risking an unclear licence status. For the agencies, these tools do two things. First, they keep enterprise clients on platform, discouraging them from taking a downloaded file to an external generator for modifications. Second, they create new metered actions, each colour swap or background removal is an event that can be priced into higher subscription tiers.

Behind the scenes, both Getty and Shutterstock are testing deeper capabilities such as generating additional poses of the same model or expanding a photo's background to fit unusual aspect ratios. The aim is simple: reduce the need for clients to commission fresh photography while still charging for the controlled, legally safe environment that an open generator cannot match.

The Future: Trusted Providers or Irrelevance?

Looking ahead, Shutterstock and Getty's strategies reflect clear awareness of these market shifts. The merged entity increasingly positions itself not merely as an image vendor but as a secure ecosystem ensuring authenticity, accuracy, and legal indemnification. They are placing a bet that regulatory and reputational fears around generative AI will preserve enough of their core market, at least for the next few years.

Yet fundamental questions remain. Will AI's relentless improvement overcome the trust advantage of real photography? Can stock agencies keep monetising licensing deals indefinitely once generative AI is sufficiently trained? Could a scenario arise where nearly every image a business needs is easier and cheaper to generate than to license from an archive?

Ultimately, the combined Getty-Shutterstock company is attempting a careful balancing act. It must protect legacy markets reliant on traditional photography, monetise emerging opportunities from AI licensing, and rapidly adapt to new content workflows. Their survival likely rests on their ability to evolve into trusted providers of verified, authenticated visuals. If not, the very concept of stock photography as we've known it could become increasingly irrelevant, replaced by a future where images are no longer purchased, but generated instantly and endlessly.