By Adam Pease
Getty Images Foreshadows Legal Battle Over AI Content
Recently, Getty Images, a large provider of stock photographs for businesses, as well as Unsplash, another stock content provider, announced its decision to ban the sale of AI-generated images from their platform. With the rise of text-to-image generative content models like Stable Diffusion and DALLE-2, the Internet has been flooded with new, generated images, and the stock photo industry is no exception. This blog discusses the decision and its implications for the future of the generative content market.
Related: Watch Aragon Research’s “What Is Generative Content?” Webinar On Demand
Why Did Getty Images Ban AI Content?
In an interview with The Verge, Craig Peters, CEO of Getty Images described “real concerns with respect to the copyright of outputs from these models and unaddressed rights issues with respect to the imagery.”
In other words, Peters voiced a concern that has been shared by many in the digital arts community following the rise of text-to-image models. While new AI models do not actually use copyrighted works to produce their results, they are trained on large datasets of images—some public domain and some copyrighted—that informs their understanding of visual concepts.
In that sense, it makes sense that legacy image providers like Getty are preparing the ground for a legal challenge to generative content. Their business model depends on being able to offer a curated selection of high-quality images, which stand out from those any average person might be able to shoot on their personal camera.
Tools like Stable Diffusion threaten to upset this monopoly by giving small business owners the tools to quickly generate the webpage or marketing content they need without paying the often-hefty licensing fees associated with stock content. It is an omen of the way AI models will soon disrupt a wide variety of creative industries.
The Legal Implications of Generative Content
Aragon Research expects that stories like this are just the first in a long series of forthcoming anecdotes that pit legacy content providers against the new realities of generative content.
Generative content has scarcely scratched the surface of its development as a field, and already it is raising powerful questions about the future of copyright law. When AI-generated content comes to the music industry, the film industry, and begins to automate human creative labor in fields beyond these early cases, we can expect to see the legal challenges mount.
But is it even possible to put the genie back in the bottle? In the interview, Peters points out that Getty is at pains to develop content filters that will automatically detect if AI content has been uploaded to its platform. And yet, as the quality of generated images continues to rise, often dipping into the territory of the photorealistic, it seems unlikely that any filter will be able to adequately discriminate between images.
It may be that these new legal challenges will simply be too difficult to enforce.
Bottom Line
The backlash against AI content by stock photo providers suggests that there will be a rocky road ahead for the adoption of generative content, as legacy providers raise legal challenges to automation.
Still, Aragon expects other providers will adapt by moving in the other direction, transforming their businesses to leverage AI for their existing customer base. Only time will tell which approach wins out.