OpenAI brings a visual ad test into ChatGPT image generation
On October 5, 2026 OpenAI announced a new visual format for ChatGPT Ads. It will first run while Free and Go users generate pictures in ChatGPT, with each ad labeled and kept apart from the user's own image. The test opens in October in the US with a first group of advertisers, alongside a wider set of measurement partners.
Ads extend from below the answer into the moment of image creation
OpenAI's October 5 announcement bundles three changes for ChatGPT: a new visual ad format, more measurement tools and partners, and new ways to judge brand suitability. By the company's own count, ChatGPT now reaches 1.2 billion people every week. Covering the news a day later, Search Engine Journal pointed out that ChatGPT ads have so far sat beneath responses, whereas this format also brings advertising into the image generation experience.
The idea is to let people picture how a product or service might fit into their lives. According to OpenAI, the ads will be built from images of product inspiration, of the product in use, or of the experiences it enables. That gives a user room to find a new brand, see what sets an offer apart and make a buying decision, and it gives advertisers a way to tell their story visually.
A narrow test, paired with a long list of measurement partners
At first the format appears only during image generation in ChatGPT, and only for people on the Free and Go plans. Every ad carries a clear label and does not blend into the picture the user is making. OpenAI repeats the principle it applies to its other formats, that "advertising does not influence the answers ChatGPT provides". Testing begins in October in the US with an initial set of advertisers. Search Engine Journal notes that OpenAI has not yet shared targeting, pricing, creative requirements or reporting details for the format, and has given no timeline for a wider rollout.
On measurement, integrations with Hightouch, Tealium and LiveRamp help advertisers pass conversion data from the systems they already run into ChatGPT Ads. For Conversions API, reporting and click attribution, OpenAI now works with AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge and Tenjin, while Fospha, Measured and INCRMNTAL are listed as full-funnel and advanced measurement partners. For incrementality, OpenAI has teamed up with Haus, Measured and WorkMagic to run geo-based experiments, work the company describes as still early. On brand safety, Negative Phrases are now open to qualifying advertisers with brand-specific placement rules, and suitability evaluation pilots are under way with DoubleVerify and Integral Ad Science, run without giving either partner access to private user conversations.
Visual creative and provable measurement arrive together
People use ChatGPT in different ways. As Search Engine Journal puts it, some research purchases while others create images or work with files, so placing ads in image generation extends the channel beyond the chat window. The same outlet adds that the gap OpenAI keeps between an ad and the generated picture is worth watching once testing starts. The early partner numbers OpenAI shared are striking. DV Rockerbox found that, for WeightWatchers, the attributed acquisition cost from ChatGPT Ads came in 15.3% below its blended paid-search benchmark. WorkMagic measured statistically significant lift for the wellness brand Dose, where 67% of incremental purchases came from net-new customers. Triple Whale reported that 93% of Portland Leather's visitors arriving from ChatGPT Ads were new.
Those figures deserve caution. Each one is reported by a partner and each measures a different outcome; Search Engine Journal warns against treating them as general benchmarks for ChatGPT Ads. The more durable change is the plumbing: advertisers can now set ChatGPT campaigns side by side with the attribution and data tools they already trust. We covered the channel's automated bidding and view-through conversions on August 25, and this announcement continues along that line.
The sources do not mention Türkiye, so this section is our view
Neither source contains any information about Türkiye. The visual ad test is explicitly framed around the US and a limited group of advertisers, so what follows is UNALSOFT's interpretation, not reported fact.
The open questions, one by one: when the visual format might reach Türkiye or any market outside the US, whether Turkish advertisers can join the test through ads.openai.com, whether ads would appear on image requests written in Turkish, and how the partner measurement integrations would behave for accounts based in Türkiye. Our reading: the attribution partners OpenAI lists span web and app, which means the channel is being built to plug into measurement setups advertisers already run. Collecting clean, consistent conversion data now is what makes a first test measurable whenever the channel widens.
The UNALSOFT view
What interests us most here is not the format but the measurement. A visual ad can show a product in the moment it is used, which opens real creative room, yet the reason to move budget into a new channel is being able to prove the result with the tools you already have. In practice that means defined conversion events, an attribution tool that is set up correctly, and product imagery that shows the context of use. That is why, in our ad management work, building the measurement layer comes before testing any new channel. Since the test currently runs only in the US, this article proposes no timeline for Türkiye.
Will you be able to measure the result when a new ad channel opens?
A short conversation is enough to get your conversion tracking, attribution setup and product imagery ready for new channels.