OpenAI adds visual ads to ChatGPT image generation

OpenAI says it will begin testing visual ads during ChatGPT image generation later this month, initially with a group of US advertisers. The ads will be labeled and displayed separately from the generated image. OpenAI says advertising does not influence ChatGPT’s answers.

Visual ads enter the image-generation flow

The format lets advertisers show product inspiration, use cases or experiences through imagery alongside image-generation interactions. This is a new placement within OpenAI’s ad product, rather than an announcement that ads will be inserted into generated images themselves. The initial test is limited to the United States and selected advertisers; OpenAI has not announced a broader launch date.

TechCrunch independently reports the same rollout timing and placement, and describes it as an expansion of OpenAI’s existing ad business. Its report also notes the initial US-only test. That distinction matters: the announcement describes a test, not general availability to advertisers worldwide.

Measurement and attribution expand

OpenAI is adding integrations that let advertisers send conversion data from systems including Hightouch, Tealium and LiveRamp. It also lists attribution partners spanning web and app campaigns, among them AppsFlyer, Adjust, Branch and Triple Whale, alongside additional reporting and full-funnel measurement firms.

The company says it is exploring geo-based incrementality experiments with Haus, Measured and WorkMagic. Those experiments aim to estimate whether ads cause additional activity, rather than merely receiving credit for conversions that might have happened anyway. OpenAI presents early partner results, including a reported cost-per-acquisition comparison and new-customer figures; these are partner-reported outcomes, not independent comparative tests.

Brand-suitability pilots are still developing

OpenAI says it is developing evaluation pilots with DoubleVerify and Integral Ad Science. The stated approach is for independent partners to assess placement safeguards in a controlled environment without accessing private conversations. The announcement also describes existing placement guardrails and a negative-phrase control for qualifying advertisers.

These are advertiser-facing controls and evaluation efforts, not a public audit of every placement. The post does not provide a launch date for the pilots or detailed performance results from them. OpenAI’s statements that answers remain independent and conversations private describe its policies; the announced test will give users and advertisers a concrete rollout to observe.

What advertisers and users can expect

Businesses can register through OpenAI’s advertising portal, while the new visual format starts as a limited US test later in October 2026. The practical change is a richer ad placement within a high-intent activity—image creation—alongside more measurement infrastructure. Whether the format expands, and how its safeguards perform at scale, remain open questions after the initial test begins.

Sources: OpenAI’s announcement · TechCrunch’s report