GPT-5.4 Is Here — And It's Built for the Office, Not the Chatbox
OpenAI just released GPT-5.4, and for once the positioning is unusually specific: this is a model designed for professional work. Not general...
The visual internet just got a little more complicated for the companies that have been hoping nobody would notice where their training data came from.
On June 21, 2026, Getty Images announced a multi-year display agreement with OpenAI, bringing its licensed content libraries into ChatGPT's search and discovery experiences. When ChatGPT surfaces visual content in response to a search query, it can now draw from Getty's catalog — content that is licensed, attributed, and commercially clean.
This is a display partnership, not a training deal. The distinction is worth holding onto because the two are often conflated and carry very different implications.
Getty's images will appear within ChatGPT when the product surfaces visual responses to user queries. The content is licensed, which means the photographers and creators whose work appears are operating under an agreement, not having their work used without consent or compensation. What the agreement does not cover, at least as publicly disclosed, is any use of Getty's library to train or fine-tune OpenAI's models.
Getty brings significant scale to this. Its combined brands — Getty Images, iStock, and Unsplash — serve customers in nearly every country and cover more than 160,000 news, sports, and entertainment events annually. The archival depth alone is unusual: millions of images dating to the earliest days of photography.
Getty is not a neutral party in the AI image debate. The company is pursuing ongoing legal action against AI companies for using its content in model training without licensing agreements. That context makes this partnership a statement as much as a business deal.
The model Getty is building — licensing content for specific, defined uses inside AI products — is a direct argument that the industry can operate differently. If AI companies can display licensed visual content with proper attribution and compensation structures, the argument that scraping was the only viable path becomes harder to sustain.
For marketers, this signals something important about where visual content in AI-generated responses is headed. The free-for-all era of AI pulling whatever images were technically accessible is closing, at least at the major platform level. What replaces it looks more like the licensing infrastructure that has governed stock photography for decades, now applied to AI search interfaces.
The practical implication for brand and content teams is not immediate, but the directional signal is clear. Visual content appearing inside AI search results will increasingly come from licensed libraries with defined usage rights. That is good news for brands that have built their creative workflows around licensed content and bad news for anyone still operating on the assumption that AI-generated or scraped visuals carry no compliance risk.
A few things worth thinking through now. First, if your team uses AI-generated images in commercial work, the question of what those images were trained on matters legally and reputationally. Getty's own generative AI tools are trained on permissioned content with built-in indemnification — a standard not universal across the market. Second, as ChatGPT becomes a more significant discovery surface, the images that appear alongside search results will shape brand perception in ways that are not yet fully mapped. Understanding which visual libraries power those responses, and under what terms, is reasonable to know.
The broader shift is toward an AI ecosystem where content provenance is tracked and compensated. The companies building governance structures around that now, rather than waiting for regulation to force the issue, will be in a stronger position when the rules are fully in place.
At Winsome Marketing, we work with growth teams navigating exactly these questions — how to build content and creative operations that hold up as AI becomes a primary channel for distribution and discovery. If you want to think through what that looks like for your brand, our team is a good place to start.
OpenAI just released GPT-5.4, and for once the positioning is unusually specific: this is a model designed for professional work. Not general...
Welcome to the age of cognitive castes. OpenAI just rolled out what they're calling "thinking effort" levels in ChatGPT—a euphemistic slider that...
OpenAI is testing a "Memory Search" feature for ChatGPT that lets users query stored information directly instead of scrolling through an...