News & analysis
OpenAI Tests Visual Ads Inside ChatGPT Image Generation
OpenAI is testing a new visual ad format in ChatGPT that will first appear during image generation, starting later this month with an initial group of U.S. advertisers.
By Roomer Editorial · · 5 min read
What is OpenAI testing?
OpenAI is piloting an ad format that shows up while someone is generating an image in ChatGPT. The test opens later this month to an initial group of U.S. advertisers. OpenAI says these ads will use images to surface product inspiration, usage, or experiences tied to what's being advertised.
Up to now, ChatGPT Ads have only appeared below chat responses, sitting inside the text conversation itself. Moving into image generation puts ads in front of a different kind of ChatGPT session, one where someone is already building or visualizing something rather than reading an answer.
OpenAI frames this as another way to reach more of its user base. The company says ChatGPT reaches 1.2 billion people weekly, and that those people use the product differently: some ask questions, some research purchases, others generate images or work with files. Bringing ads into image generation opens advertising to more of those experiences rather than relying only on the traditional chat interface.
How will the ads stay separate from image generation?
OpenAI says the new ads will be clearly labeled and kept visually separate from the image a user is actually creating. The company also states that advertising will not influence ChatGPT's responses.
That's the official line. Whether it holds up once the format goes live is a fair question. An ad that pops up mid-image-creation might blend into the experience more easily than one stacked under a block of text, labeling aside. How much separation OpenAI actually creates between the two will be worth watching once the test begins.
OpenAI hasn't published details on targeting, pricing, creative specs, or reporting for this format yet. That's consistent with a test running on a limited group of advertisers rather than a public launch.
What measurement and brand safety tools come with this?
Alongside the visual ad test, OpenAI is rolling out measurement and brand safety infrastructure for ChatGPT Ads generally, not just for the new format.
On measurement, OpenAI is adding integrations with Hightouch, Tealium, and LiveRamp, letting advertisers send conversion data from their existing systems into ChatGPT Ads. The company also supports a list of attribution partners, including AppsFlyer, Triple Whale, Adjust, DV Rockerbox, and Northbeam, covering capabilities such as reporting, click attribution, and Conversions API integrations.
OpenAI is separately testing incrementality measurement with Haus, Measured, and WorkMagic, using geo-based experiments designed to measure whether ChatGPT Ads caused additional business results, rather than simply getting credit for sales that would have happened regardless.
The company shared a handful of early results from these partners. DV Rockerbox found WeightWatchers' attributed cost per acquisition came in 15.3% below its blended paid search benchmark. WorkMagic reported a statistically significant incremental lift in purchases for Dose. Triple Whale found that 93% of Portland Leather's visitors from ChatGPT Ads were new.
These figures measure different outcomes for different advertisers, and OpenAI itself describes the incrementality work as early stage. None of this should be read as a general ChatGPT Ads benchmark.
On brand safety, OpenAI introduced a feature called Negative Phrases, which lets qualifying advertisers with their own placement requirements keep ads away from specific topics or contexts tied to their brand policies. The company also has existing guardrails meant to assess whether a given conversation is appropriate for advertising at all, intended to keep ads away from sensitive, emotionally vulnerable, or otherwise unsuitable contexts. On top of that, OpenAI is developing brand suitability pilots with DoubleVerify and Integral Ad Science, who will evaluate how OpenAI's brand safety standards are applied, without receiving access to private user conversations.
Who has access right now?
Very few advertisers. OpenAI is opening the visual ad test to a limited initial group in the U.S., with no stated timeline for a broader rollout. For most advertisers, this is a format to watch rather than something to plan around today.
If access expands, advertisers will need to look past click-through rates or whether the new ads simply outperform existing ChatGPT placements. The open question, per the source, is whether visual ads reach people or influence behavior that advertisers weren't already capturing through other channels. That's where the new attribution integrations and early incrementality testing could matter, by helping advertisers connect ChatGPT Ads to conversions and then check whether those conversions would have happened anyway.
Roomer Editorial's take
Roomer Editorial's view: the ad format itself is a small story. The bigger one is that OpenAI is building the standard advertising stack of a mature platform, attribution partners, incrementality testing, brand safety review, around a product that already touches more than a billion people a week. That is infrastructure built by a company planning to keep the commercial relationship with its users close, not pass it along to someone else's rails.
For a travel creator, the useful question isn't whether ChatGPT ads look native or intrusive next to a generated image. It's where the booking actually happens once someone acts on something they saw there. If a traveler asks ChatGPT to generate a mood board for a Lisbon trip and a visual ad shows up beside it, that's travel intent getting captured inside OpenAI's product, on OpenAI's terms, measured by OpenAI's attribution partners, not yours.
Roomer Editorial's view is that this is another argument for keeping the booking step on infrastructure you control.
Up to now, ChatGPT Ads have only appeared below chat responses, sitting inside the text conversation itself.
A page on your own site, built from your own videos, with bookable links running through Vio, Stay22, GetYourGuide, Viator, or Klook, keeps the commission with you no matter what ad formats other platforms test around their own traffic. You can't do anything about what ad shows up next to someone's ChatGPT session. You can control what happens the moment they land on your page instead, and you keep 100% of the commission on bookings made there.
None of this means ignore where ChatGPT traffic to your content comes from. It means not building your booking funnel on top of a platform that's actively testing new ways to insert itself between a traveler's intent and your commission.
Action: Check where your highest-intent travel content currently sends people to book, and move anything routing through a platform you don't control to a bookable page on your own site instead.
Source: Search Engine Journal — OpenAI To Test Visual Ads In ChatGPT Image Generation
