Learn how marketers can unlock the power of Audio to drive impact at Adweek Crack the Code: Audio's Next Play, October 5. Register for free.
“Man and woman stand at a crossroad.” So reads the description for a photo on Getty Images, a go-to provider for stock imagery. With Shutterstock, they own or license hundreds of millions of photos and illustrations, sourced from millions of contributors, to agencies and their clients for use in ads, presentations and other materials. If you’ve browsed a website or read a magazine, you’ve seen licensed stock.
Or at least that’s been the case. Like many industries today, the stock photo world is undergoing ground-shifting changes from generative AI—particularly text-to-image generators like Midjourney, DALL-E 2 and Stable Diffusion. The technology has brought new entrants into the competitive landscape, forcing the traditional industry to reorient itself to protect its business.
Yet for advertisers and agencies, these changes present an opportunity to assert their own interests—with both traditional stock photo licensors and the emerging GenAI companies.
GenAI imagery threatens some artists and photographers more than others. For well-known visual artists, these tools likely won’t hurt the market for their work. If you want Andrew Wyeth’s 1948 painting “Christina’s World,” for example, you want the real thing.
But for businesses trading in more generic images like Getty and Shutterstock, GenAI tools present a competitive threat. These tools can convincingly generate the type of stock imagery common in many ads, like “Man and Woman Stand at a Crossroad.” In economic terms, AI-generated images are a closer substitute good to human-created stock: Marketers can use either product for similar ends.
GenAI tools’ price and flexibility make them particularly threatening to traditional stock licensors. These companies often sell licenses a la carte—a high-res version of “Crossroad” goes for $499. But GenAI tools let users obtain near-unlimited photos for a modest monthly fee—for Midjourney, between $10 and $120 a month. And while stock services offer a collection of ready-made photos, GenAI allows users to tailor images to suit their exact needs.
Of course, the resulting images aren’t perfect substitutes for photos created by humans. Much AI-generated art looks like, well, AI-generated art. At its worst, it’s grotesque—hands with too many fingers, faces out of horror movies. But even at its best, AI art has shortcomings, especially for advertisers and agencies eager to incorporate AI into their campaigns.
AI-made material isn’t protected by copyright, at least under U.S. Copyright Office policy and a recent federal court decision. It also typically lacks the non-infringement warranties that most commercial licensees demand for licensed assets—warranties that Getty and other reputable stock providers offer. GenAI companies’ failure to offer these warranties is particularly troubling given the recent spate of lawsuits involving AI-generated art. If a marketer gets sued for AI-generated imagery used in an ad, they may find little help from the provider of the technology; the legal claim—and any ensuing costs and damages—may land in the marketer’s lap.
Even so, for all the technology’s legal concerns and occasionally disturbing outputs, GenAI tools may still meet a marketing exec’s immediate need: illustrating a scene for a budget-constrained ad campaign.
At a crossroads
With new entrants encroaching upon their territory, companies like Shutterstock and Getty face a critical choice: to beat ‘em or join ‘em.
Shutterstock is doing the latter. In July, it announced a six-year deal with OpenAI—maker of the text-to-image generator DALL-E 2 and the wildly popular chatbot ChatGPT. Under the agreement, OpenAI can train its models on Shutterstock’s content; in return, Shutterstock can use OpenAI’s tech for its own AI offerings. Shutterstock has also inked partnerships with Meta and LG.
Getty, too, is developing an AI model with Nvidia. But, simultaneously, it’s taken to the courts. In February, the company filed a federal suit in Delaware against Stability AI, creator of Stable Diffusion. Among other claims, Getty alleges that Stability unlawfully scraped over 12 million Getty photos and metadata to train its GenAI model, and that the model’s outputs infringe Getty’s IP.
The companies’ two approaches—litigation and deals—parallel how another entrenched industry responded to a disruptive new technology. When Napster released its peer-to-peer file-sharing software at the turn of the millennium, most major record labels sued. Meanwhile, German entertainment conglomerate Bertelsmann partnered with the startup to create a paid membership service. The litigators ultimately prevailed: After major court losses, Napster filed for bankruptcy, while Bertelsmann shelled out millions to the labels in settlements.
Whether lawsuits by Getty and others will derail today’s GenAI companies remains to be seen. But however those cases end, GenAI’s effect on the stock photo industry will be permanent. Getty and Shutterstock are both embracing the new technology by developing their own generative AI models, as is Adobe, who debuted its tool Firefly in March.
By doing so, these companies aren’t simply expanding into new offerings; they’re potentially changing their business models. Shutterstock and Getty have promised to compensate their contributors when their images are used to train models. But once those models are trained, the value of new human-generated stock photography will likely decline, with much of the demand being met by GenAI.
That shift risks putting stock licensors at odds with their own contributing photographers, whose works have long been these companies’ main asset. These contributors have little leverage to fight the competitive reorientation. While stock licensors give their contributors the ability to opt out of using their works to train AI models, any one photographer’s decision to do so will have minimal effect on the models, which learn on millions upon millions of images.
The opportunity for marketers
Generative AI is dramatically altering the traditional stock photo industry, but for advertisers and agencies that regularly license stock, this volatility offers an opportunity to assert their own interests with not just stock licensors but also with GenAI companies.
Where once marketers had to rely on a handful of companies for their stock photography, they can now employ tools like Midjourney to source their images. The emergence of these new vendors gives marketers a chance to press their current stock licensors for competitive products and more favorable licensing terms.
Marketers also have an opportunity to demand better offerings from GenAI companies. Because most GenAI tools still lack non-infringement warranties, many advertisers and agencies may hesitate to use them for their campaigns. Companies familiar with the often massive damages awarded in copyright infringement lawsuits may happily pay a premium for the contractual protections that Getty, Shutterstock and other established players offer. That reality gives marketers grounds to ask for similar protections from providers of text-to-image GenAI technology.
Only time will tell how generative AI will affect the ubiquitous stock photo. But as existing and new industries alike rush to protect their interests in this new competitive world, prudent marketers won’t just watch from a distance; they will ensure their own interests are protected too.