AI Display Deals Need Usage Evidence: The Publisher Meter
Getty Images’ agreement with OpenAI will be read by many as a stock-market story.
That is too narrow.
The more important signal is not what happened to Getty’s share price. It is what the deal reveals about the next phase of AI licensing.
Getty described the agreement as a multi-year display partnership. Its licensed content libraries will appear across OpenAI search and discovery experiences within ChatGPT.
That matters because the value is no longer only created before an AI answer is generated. It is created inside the answer experience itself.
And once value moves into the AI interface, one question becomes unavoidable.
Who holds the meter?
In short
An AI display deal is different from an AI training deal.
A training deal concerns the use of content to build, train, fine-tune or improve a model. A display deal concerns content that appears, supports or enriches the user-facing AI experience — for example, an image, citation, source, excerpt, retrieved item or licensed reference displayed inside an answer.
That distinction changes the economics for publishers.
In display deals, value may be created at the moment of user interaction. The content may increase trust, context, answer quality and commercial relevance. But that value may not produce a click, a pageview or a visible signal on the publisher’s side.
At Citations Logic, we use the term usage evidence to describe publisher-side records that make AI content use visible, attributable and commercially actionable.
A platform report is useful. But it is not the same as independent evidence.
A platform report is a statement. A publisher-side usage record is leverage.
Display is not training
The Getty/OpenAI agreement is explicitly positioned around display.
Getty announced that its licensed content libraries will appear across OpenAI search and discovery experiences within ChatGPT. The agreement enables Getty content to be used for display inside ChatGPT visual responses.
That is not the same commercial category as model training.
Training deals focus on whether content can be used to develop or improve a model. The negotiation usually revolves around permission, scope, compensation, restrictions and lawful use.
Display deals focus on what users see or experience inside an AI product.
That is a different value moment.
In a traditional web model, a publisher could often observe value indirectly: impressions, clicks, referrals, subscriptions, downstream conversions.
In an AI answer model, the user may receive the value without leaving the interface.
The content is used.
The publisher may not see the use.
That is the shift.
This is why AI usage evidence for publishers is becoming a commercial issue, not only a technical one.
The real question is not “do we get reporting?”
Usage reporting is becoming a standard answer to this problem.
Microsoft’s Publisher Content Marketplace is a clear example. Microsoft says PCM is designed to create a transparent value exchange between publishers and AI builders, with publishers defining licensing and usage terms while AI builders license content for grounding scenarios.
Microsoft also says PCM provides usage-based reporting so publishers can understand how content has been valued and where it can create future value.
That is a meaningful step.
But it does not remove the core asymmetry.
In many AI licensing models, the usage report is produced by the platform running the AI experience.
The publisher receives the number.
The publisher does not produce it.
That difference is not administrative. It is commercial.
When the renewal conversation starts, the party that controls the logs controls the negotiation frame.
The publisher can ask questions.
The platform holds the meter.
That is where proof of AI content usage becomes strategically different from ordinary reporting.
Usage reporting vs. usage evidence
The distinction matters enough to name it clearly.
Question | Usage reporting | Usage evidence |
|---|---|---|
Who produces the record? | Usually the AI platform or marketplace | The publisher, rights holder or independent evidence layer |
What does it provide? | Reported usage metrics | Verifiable usage records |
What is it useful for? | Visibility, dashboards, commercial summaries | Renewal, audit, dispute support, contributor compensation |
What is the risk? | Dependence on the platform’s definitions and logs | Requires instrumentation, governance and rights mapping |
Strategic value | Better than opacity | Stronger negotiating position |
Usage reporting answers: “What did the platform say happened?”
Usage evidence answers: “What can the publisher prove happened?”
That is the commercial fault line.
You cannot price what you cannot count
Flat-fee licensing made sense when AI usage was hard to observe.
If no one can reliably count how often a catalogue is used, a lump-sum agreement may be the only practical structure.
But once usage becomes countable, it becomes negotiable.
That is why display deals matter.
They move the licensing discussion from access to observed value.
Which asset appeared in an AI answer?
Was it displayed, cited, summarized, retrieved or used for grounding?
Was attribution visible to the user?
Was the source shown directly, or did it only support the response behind the interface?
Can the event be linked back to a work, collection, contributor, right, territory or commercial category?
Without those records, an AI licensing deal can remain a black box.
The publisher may know its catalogue was included in an agreement. It may not know which parts of the catalogue created value, how often they were used, or what should change at renewal.
That is a weak position.
“Trust us, it was used a lot” is not a pricing model.
That is also why the difference between retrieved is not cited matters. A source can support an answer without becoming visible to the user.
The side holding the logs holds the renewal
The first AI licensing deal often starts with access.
Can the AI company use the content?
For what purpose?
Under what restrictions?
At what price?
The second deal is different.
The second deal asks what happened.
How often was the content used?
Which assets mattered?
Which sources improved answer quality?
Which contributors should be compensated?
Which rights were activated?
Which terms should change?
If the publisher cannot answer those questions independently, it enters renewal with a request.
If it can produce usage evidence, it enters renewal with a record.
That difference will shape the next phase of AI content licensing.
This is where recurring AI licensing revenue becomes harder to defend without a measurement layer.
Access may justify the first deal.
Evidence will decide the second.
Why this matters beyond Getty Images
Getty is a visual content company. Its catalogue is not the same as a scientific journal archive, a legal database, a medical reference library or an educational publisher’s curriculum.
But the pattern applies across authoritative content.
AI systems increasingly need trusted sources to produce better answers: images, news, scientific evidence, legal references, medical information, educational materials, standards and specialist databases.
Regulators are moving in the same direction.
In June 2026, the UK Competition and Markets Authority required Google to give publishers more control over the use of their content in AI search features and to ensure proper attribution with clear links in AI-generated search results.
That confirms the direction of travel.
Attribution, control and bargaining power are no longer side issues.
They are becoming market structure issues.
But attribution alone is not enough.
A link tells the user where something came from.
Usage evidence tells the publisher what happened, how often, under which conditions, and with which commercial consequence.
That is why AI content attribution for publishers must be connected to usage records, not treated as a cosmetic label.
The objection: platform reporting may be enough
There is a reasonable objection.
If a major AI platform provides usage reporting, why should publishers need another evidence layer?
Because reporting and auditability are not the same thing.
Platform reporting can be useful. It can give publishers visibility they did not previously have. It can support dashboards, revenue-sharing models and internal analysis.
But a report controlled by the counterparty has limits.
Who defines what counts as a use?
Does retrieval count if the source is not displayed?
Does display count if there is no click?
Does grounding count if the content influenced the answer but was not cited?
Can the publisher verify the logs?
Can contributor compensation be traced to actual usage events?
Can the records support renewal, audit or dispute resolution?
If those questions are unanswered, reporting improves visibility without solving dependence.
Better visibility is not the same as negotiating leverage.
What publishers should require in AI display deals
Before signing or renewing AI display deals, publishers should require clarity on measurement.
The agreement should distinguish between training, retrieval, grounding, summarization, citation, display and downstream reuse.
It should define what counts as a payable use.
It should specify whether usage can be linked to specific assets, collections, contributors, rights categories and territories.
It should explain who controls the logs.
It should define audit rights.
It should state whether usage evidence affects renewal pricing.
It should explain how contributors, authors, photographers, journalists or content partners are compensated.
These are not operational details.
They are the commercial architecture of AI licensing.
Access is only the first asset
The Getty/OpenAI deal will probably continue to be discussed as a market signal.
That is fair.
But publishers should read it as something more precise: a measurement signal.
AI licensing is moving beyond access. Display deals put publisher value inside the AI answer itself. That makes usage evidence essential.
A publisher that owns a valuable catalogue but cannot prove how it is used inside AI systems is still negotiating from partial visibility.
A publisher that can produce its own usage record is in a different position.
That is the side of the problem Citations Logic focuses on: helping authoritative content owners make AI usage visible, attributable and commercially actionable from their side of the table.
Because in the AI content economy, the decisive question may soon be less:
“Was our content included?”
And more:
“Can we prove how it was used?”
Frequently asked questions
What is an AI display deal?
An AI display deal allows licensed content to appear inside an AI user experience, such as a search answer, discovery surface, visual response, citation module or retrieved source shown to the user.
How is a display deal different from a training deal?
A training deal concerns the use of content to build, train, fine-tune or improve an AI model. A display deal concerns content used or shown in the user-facing AI experience.
Why do AI display deals need usage evidence?
Because display value may happen inside the AI interface without generating clicks, pageviews or publisher-side signals. Usage evidence helps publishers support renewal pricing, audit rights and contributor compensation.
Is usage reporting enough for publishers?
Usage reporting is useful, but it may still depend on the platform’s definitions, logs and interface. Publisher-side usage evidence gives content owners a stronger basis for renewal, audit and negotiation.
Continue the evidence chain
AI Usage Evidence for Publishers
One-Time vs Recurring AI Licensing Revenue
AI Content Attribution for Publishers
Book an AI usage evidence assessment
Sources
Getty Images — “Getty Images Announces Display Partnership with OpenAI”
https://newsroom.gettyimages.com/en/getty-images/getty-images-announces-display-partnership-with-openai
Microsoft Advertising — “Building Toward a Sustainable Content Economy for the Agentic Web”
https://about.ads.microsoft.com/en/blog/post/february-2026/building-toward-a-sustainable-content-economy-for-the-agentic-web
UK Competition and Markets Authority — “CMA secures fairer deal for publishers and improves Google search services in UK”
https://www.gov.uk/government/news/cma-secures-fairer-deal-for-publishers-and-improves-google-search-services-in-uk
Open Markets Institute — “Same Gatekeepers, New Tollbooths: Mapping the AI Content Licensing Market”
https://www.openmarketsinstitute.org/publications/report-mapping-the-ai-content-licensing-market
Nieman Lab — “The emerging AI content licensing market puts news publishers in a double bind, a new report warns”
https://www.niemanlab.org/2026/05/the-emerging-ai-content-licensing-market-puts-news-publishers-in-a-double-bind-a-new-report-warns/