Loading page…
The US trial will keep ads separate from generated images, OpenAI says. New measurement integrations and safety evaluations address advertisers’ questions about results and placement.
Listen to this story
OpenAI plans to test visual ads in the US with a limited group of advertisers later in October 2026, placing them alongside ChatGPT image-generation sessions rather than inside generated artwork. The test is paired with measurement integrations and early geographic experiments designed to distinguish incremental business impact from ordinary conversion attribution. OpenAI says ads will be labeled, will not influence answers, and will be screened for suitable placements; its brand-safety pilots are designed to operate without access to private conversations.
OpenAI cited partner findings including a 15.3% lower attributed acquisition cost than WeightWatchers’ blended paid-search benchmark, 67% of Dose’s incremental purchases coming from new customers, and 93% new visitors for Portland Leather.
Hightouch, Tealium, and LiveRamp can connect advertisers’ conversion data to ChatGPT Ads; AppsFlyer, Adjust, and Triple Whale support attribution reporting.
Haus, Measured, and WorkMagic will explore geographic experiments to test whether ads cause additional business activity, a different measure from assigning purchase credit.
People generating images in ChatGPT will soon encounter a new kind of visual advertising—but OpenAI says the ads will stay outside the images themselves. In its October 5 announcement, the company outlined a US test with an initial group of advertisers later in October 2026, alongside new tools for measuring ad results and evaluating suitable placements.
The format uses images to show product inspiration, products in use, or experiences a service makes possible. OpenAI’s pitch is that these visuals can help people imagine an offering in their own lives while giving advertisers more room to explain it.
The initial placement is during image generation, not inside the resulting artwork. OpenAI says ads will be clearly labeled and will not influence ChatGPT’s answers. This is a planned test with a limited advertiser group, rather than a completed general rollout.
For advertisers, the accompanying measurement changes connect ChatGPT Ads to systems they already use. Integrations with Hightouch, Tealium and LiveRamp let businesses send conversion data—records of purchases or other tracked actions—from their existing systems to ChatGPT Ads, OpenAI says.
OpenAI also supports web and app attribution partners, including AppsFlyer, Adjust and Triple Whale, for reporting and connecting clicks to outcomes. A separate, early-stage effort with Haus, Measured and WorkMagic will explore geographic experiments to assess whether ChatGPT Ads cause additional business activity.
That causal question is distinct from assigning credit for a purchase or counting new visitors. OpenAI highlighted three early partner findings below. They describe different measures of ChatGPT Ads performance, not results from the visual-format test that has yet to begin.
OpenAI says its placement guardrails assess whether a conversation is appropriate for advertising, keeping ads out of emotionally vulnerable, sensitive or otherwise unsuitable contexts. Automated review, human oversight and ongoing monitoring support those safeguards. Qualifying advertisers can now use Negative Phrases for narrower placement requirements tied to their own policies.
The company is developing evaluation pilots with DoubleVerify and Integral Ad Science to examine how its brand-safety standards are applied. Those assessments will use a controlled testing environment without access to private user conversations, OpenAI says.
Loading discussion...
Join the conversation
Explain what would make the separation convincing.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.