Getty Images Skyrockets on ChatGPT Link-Up

The Getty–OpenAI display deal is a textbook example of how AI platforms and rights holders are learning to work together in public view on distribution, while keeping the most consequential question — training rights — behind closed doors.

Key Points

Getty Images and OpenAI have signed a multi‑year display agreement that pipes Getty’s licensed image libraries directly into ChatGPT’s search and discovery experiences.^[2]
The public description is narrowly about display: Getty content appears in ChatGPT answers as rights‑cleared visuals, with no disclosed permission for training OpenAI’s models on that library.^[2]^[3]
Getty has a parallel track record of explicitly licensing both display and AI‑training uses to other partners, and of litigating unlicensed training, underscoring that training rights are a distinct, negotiable category.^[11]^[12]^[15]^[23]
For creators and customers, the deal improves provenance and legal safety of images inside ChatGPT — but it does not resolve broader transparency concerns over AI training or future reuse of visual archives.^[3]^[24]^[26]

What the Getty–OpenAI Deal Actually Does

Getty’s announcement is unusually clear on one dimension and conspicuously silent on another. In a GlobeNewswire press release, the company describes a “multi-year display agreement” under which its licensed content libraries “will appear across OpenAI search and discovery experiences within ChatGPT,” and states that the agreement “enables the use of Getty Images’ content for display within ChatGPT.”^[2] In plainer terms, when a ChatGPT response calls for a photograph or illustration, the system can show a real Getty image rather than only synthetic output.

That is not a minor embellishment. For OpenAI, direct access to one of the world’s largest commercial image archives means its answers can carry photos with clear provenance, creator credit, and a licensing chain compatible with commercial use. For Getty, the deal is a distribution play: its images, long licensed to publishers and marketers, now sit inside a widely used conversational AI interface, with the potential to drive new licensing revenue and visibility.^[1]^[7]

Market reaction underlines how investors interpret this move. Coverage in financial media and trading commentary highlights Getty’s stock surging well over 100% in pre‑market trading after the announcement, explicitly tying the jump to the OpenAI display partnership rather than to any broader change in Getty’s fundamentals.^[4]^[7] Wall Street is reading this as a structural re‑rating driven by AI distribution, not as a one‑off marketing tie‑in.

Display Rights Versus Training Rights: A Deliberate Separation

Equally important is what the release does not say. Neither Getty nor OpenAI has publicly disclosed any terms about using Getty’s library to train current or future AI models. Technology press and analyst write‑ups stress this omission, noting that the announcement “is about display, surfacing existing licensed pictures, rather than about feeding them into a model,” and that no financial terms or revenue split were made public.^[3] In other words, the only committed, on‑the‑record use is that Getty content may be displayed in ChatGPT responses.

This narrow framing is not accidental. Getty’s recent history shows it treats training rights as a separate, explicitly negotiated category. In a public statement related to its litigation against Stability AI, the company emphasized that it has “provided licenses to leading technology innovators for purposes related to training artificial intelligence systems,” making clear that such training requires its own license rather than being implied by general access to the site or catalog.^[12] Training is a distinct economic and legal use case.

That separation is reinforced by Getty’s own AI products. When it launched “Generative AI by Getty Images,” the company stressed that the tool is trained “solely from Getty Images’ vast creative library, including exclusive premium content,” and that outputs are indemnified for commercial use.^[11] Training provenance and downstream usage rights are part of the offer; they are spelled out as unique value. If Getty were granting OpenAI similar training rights on its main library, one would expect at least a hint of that fact in the marketing narrative — just as it does when advertising its own AI tools and data partnerships.^[11]^[22]

How This Fits Getty’s Broader AI Licensing Strategy

The OpenAI agreement does not emerge in isolation. Getty has been moving along two parallel tracks: suing over unlicensed training on one hand, and signing carefully scoped AI partnerships on the other. In its UK and US cases against Stability AI over the Stable Diffusion model, Getty alleged that millions of its images were scraped without authorization to train the system.^[13]^[23] While jurisdictional and evidentiary issues led Getty to withdraw core training claims in the UK and lose on secondary infringement, the litigation framed training on its catalog as “the lifeblood” of its business and something that must be licensed, not assumed.^[23]^[25]

At the same time, Getty has been increasingly willing to license display rights to AI search and assistant products. A 2025 announcement with Perplexity describes a “global multi‑year licensing agreement covering display of images from Getty Images across Perplexity’s AI‑powered search and discovery tools,” integrated via Getty’s API.^[14] Commentary around that deal — including from industry analysts — emphasized that while Perplexity could display Getty images, it did not obtain permission to train its models on that content, marking a deliberate distinction from broader data‑access deals.^[5] The OpenAI agreement uses almost identical language around “display” in search and discovery, suggesting a similar structure.^[2]^[14]

Getty’s generative‑AI collaborations reinforce the pattern. Its engagement with Clarifai, for example, is explicitly about integrating “Generative AI by Getty Images” into Clarifai’s platform, giving enterprise customers access to AI‑generated visuals that Getty presents as responsibly developed and commercially safe.^[22] Here, the training set and licensing regime are part of the product’s identity. By contrast, the ChatGPT deal is framed as distribution of pre‑existing, rights‑cleared images inside another firm’s assistant, with no suggestion that Getty’s core library is becoming raw fuel for OpenAI’s next model generation.^[2]^[3]^[22]

Why the Ambiguity Matters for Creators

For photographers and other rights holders whose work lives in Getty’s catalog, the OpenAI deal raises two very different questions: what happens on screen, and what happens under the hood. On screen, the arrangement is largely positive. Getty and OpenAI describe an experience where “every image shown is licensed and rights-cleared,” with proper creator credit and safer reuse for end users.^[6] That aligns with Getty’s own research showing that nearly 90% of consumers want transparency on AI images and that “authentic” visuals are central to trust in brands.^[26] A ChatGPT answer that includes a credited Getty photo is, from a provenance standpoint, cleaner than one with an unlabeled AI‑generated picture.

Under the hood, however, creators are left inferring rather than reading firm guarantees. Getty has demonstrated in other contexts that it can and will draw contractual lines around training: licensing to “leading technology innovators” for training AI systems on one side, and suing alleged scrapers on the other.^[12]^[15]^[23] Yet in the OpenAI release, there is no statement that training is prohibited, no affirmation that the deal is display‑only, and no description of whether OpenAI may store, transform, or analyze usage data related to those images to inform its future systems.^[2]^[3]

Given that Getty positions itself as a champion of transparency and legal safety in AI imagery — up to and including indemnifying customers of its own AI tools against copyright claims^[11]^[24] — this silence is striking. It does not prove that hidden training rights exist; more likely, based on the company’s pattern with Perplexity and its public hostility toward unlicensed training, the operational agreement focuses on display.^[5]^[11]^[14]^[19] But creators who care about how their work feeds machine‑learning systems have no direct way, from the public documents, to confirm the boundaries of this particular partnership. That gap between marketing language and technical reality is fast becoming one of the defining tensions in AI content licensing.

OpenAI secures a deal to integrate Getty Images into ChatGPT results. This partnership enhances the visual content offering within AI interactions, potentially challenging Google's dominance in search-related image features. @OpenAI #OpenAI #ChatGPT #AINews pic.twitter.com/IkKT6hz9jS

— Alesha AI (@alesha_ai) June 22, 2026

Implications for AI Platforms, Investors, and the Next Wave of Deals

For OpenAI, the Getty partnership is one more step in a broader pivot away from scraping‑first data practices toward visible, paid licensing. Media and market analysis point to a “growing list of media and content deals, mostly around licensed content appearing in ChatGPT answers,” and interpret the Getty agreement as evidence that the company is now “trying to replace scraping/legal fights with paid licensing, attribution, and direct publisher integration.”^[4] That shift is pragmatic: as courts and regulators wrestle with the legality of using copyrighted material for training, the easiest way to de‑risk some parts of the stack is to license what users actually see.

Investors are rewarding that strategy. Getty’s stock‑price spike reflects a belief that AI assistants are not just marginal new channels, but central aggregators of information where high‑quality visual inventory will command premium placement and recurring fees.^[4]^[7] If ChatGPT becomes a core discovery surface for news, shopping, and research, then appearing, with attribution, in its answers matters as much as page‑one placement once did in traditional search. That potential helps explain why a single display‑only agreement can move a stock by triple‑digit percentages in a single session.

At a system level, the Getty–OpenAI deal illustrates a broader pattern. Across the AI ecosystem, companies increasingly announce display or discovery partnerships with content owners while leaving the deeper question of training rights opaque or unaddressed. The immediate business incentives — better user experience, cleaner legal posture for visible outputs, and new revenue for rights holders — align neatly around these front‑of‑house arrangements.^[3]^[5] The harder issues of how data is ingested, how long it is retained, and what future models may learn from it remain locked in private contracts that neither side is eager to summarize in a press release.

What to Watch Next

For practitioners, rights holders, and policy makers trying to understand where this is heading, several concrete signals are worth tracking. First, how Getty content actually appears in ChatGPT: Does each image carry visible credit and licensing cues? Are there easy paths from a ChatGPT answer to a Getty licensing page, or to a photographer’s portfolio? Those choices will show how seriously OpenAI and Getty treat attribution and conversion from display to paid usage.^[3]^[14]

Second, whether Getty images inside ChatGPT remain confined to the consumer interface or also surface through developer APIs and partner integrations. Wider technical exposure typically correlates with more complex questions about caching, copying, and secondary use, which could blur the practical line between “display” and “data.”^[3] Third, how Getty structures future AI deals. If subsequent agreements with other platforms explicitly spell out the absence of training rights — as commentary around the Perplexity deal suggests — pressure will mount to clarify the OpenAI terms as well.^[5]^[14]

Finally, expect the legal landscape to continue shaping business behavior even when companies prefer to talk only about user experience. The mixed but highly visible outcome of Getty’s cases against Stability AI, along with parallel copyright suits in the United States, has already influenced how both AI developers and content owners write their contracts.^[15]^[17]^[25] As courts draw sharper lines around what counts as infringement in training and deployment, the kind of ambiguity embedded in “display only” AI partnerships will become harder to sustain. For now, the Getty–OpenAI agreement is best understood as a strategic truce on the visible surface of AI — a way to make ChatGPT’s images more legitimate and lucrative — while the underlying conflict over who gets to teach the machines with which images continues, largely, offstage.