Getty Images Signs Display Deal with OpenAI Amid Strategy of Licensing and Litigation

A Two-Pronged AI Strategy: License and Litigate

Getty's approach to AI companies over the past two years reveals a clear dual strategy: monetize its library through display-licensing deals with AI search platforms while aggressively pursuing legal action against companies it believes used its content without authorization to train generative models.

The Licensing Track: Perplexity AI (October 2024)

Before the OpenAI deal, Getty had already tested the waters with a similar arrangement. On October 31, 2024 (reported in some sources as 2025), Getty and AI search startup Perplexity announced a global multi-year licensing agreement . Under that deal, Getty supplies images via an API integration so Perplexity can display them in its AI-powered search and discovery tools, with built-in credits and links back to original sources . Like the OpenAI agreement, the Perplexity deal is a display-focused licensing arrangement, not a training-data license — significant because Perplexity does not train its own foundation models .

The Litigation Track: Stability AI (November 2025 UK Ruling)

On the other side of Getty's strategy is litigation. Getty sued Stability AI in the UK, alleging that Stability's Stable Diffusion image-generation model was trained on millions of Getty images without permission . However, the outcome was a major setback for Getty.

On November 4, 2025, the UK High Court handed down its ruling :

• Getty withdrew its primary copyright infringement claims before the final ruling, citing insufficient evidence about the specific training sources of Stable Diffusion .
• The court rejected Getty's secondary copyright infringement claim, ruling that the Stable Diffusion model does not contain or store reproductions of the copyrighted works and therefore is not an "infringing copy" .
• Getty won only a narrow trademark-related victory — the court found limited liability for outputs from certain early versions of the model that displayed Getty or iStock watermarks .

The ruling was widely interpreted as a landmark win for AI developers in the UK, establishing that an AI model's weights are not a stored copy of training data and thus cannot be the basis for secondary copyright infringement claims .

What the OpenAI Deal Means for the Industry

The OpenAI agreement is the latest and most significant expression of Getty's "license where you can, litigate where you must" playbook . Key implications include:

• ChatGPT gains a licensed, high-quality visual layer. Users searching or asking questions within ChatGPT will now see professional imagery from Getty's library alongside text responses, improving the visual richness and credibility of answers .
• Getty secures a new revenue stream and distribution channel. The deal monetizes Getty's library in the AI search space, a growing battleground for content licensing .
• The "display, not train" model becomes a template. Both the OpenAI and Perplexity deals explicitly separate display rights from training rights, offering a blueprint for how stock content providers can negotiate with AI companies .
• It strengthens OpenAI's copyright-compliance posture. By securing a license for professional visual content, OpenAI can point to this deal as evidence of its commitment to licensing and attribution, reducing legal risk .

A Strategy of Contrasts

Getty's AI strategy is one of stark contrasts: it is willing to partner with AI search companies on display licensing, but it is also willing to spend years in court fighting generative AI companies. The outcomes so far are mixed:

The thread connecting these moves is clear: Getty is building a business model that monetizes its library in the AI search era through display licensing, while using litigation to fight for compensation — and clearer legal rules — around unauthorized training uses. Whether this two-pronged strategy will ultimately succeed remains an open question, but the OpenAI deal suggests that for AI search products at least, licensed display content is becoming the new standard.