OpenAI Is Testing Visual Ads Inside ChatGPT’s Image Generator

OpenAI has begun testing a visual ad format inside ChatGPT's image generation feature, rolling it out to Free and Go plan users in the United States. It's the company's first foray into traditional advertising, and it marks a notable shift in how one of the world's most-used AI products is funded.

How the Ads Work

The placement is specific: ads appear while an image is generating, not inside the generated image itself. OpenAI says in its help documentation that the ads are clearly labeled, kept visually separate from the output, and do not affect ChatGPT's answers or the generated image in any way. The format is designed around product inspiration — think of it as a "picture this" placement where a brand can show how its product fits a context the user is already imagining.

A multi-image variant is also being tested, where multiple branded visuals cycle during generation. The ads are currently limited to a small group of advertisers in a closed pilot, with the launch timed to Advertising Week 2026.

Measurement and Brand Safety

Alongside the ad format, OpenAI is rolling out measurement and brand-safety tools for advertisers. Pilots with DoubleVerify and Integral Ad Science are underway, giving brands the third-party verification they need before committing significant spend to a new platform. Without those guardrails, most major advertisers wouldn't touch a new placement regardless of reach.

This matters because OpenAI is trying to attract brand budgets that currently flow to Google and Meta, and those advertisers have compliance requirements baked into their media contracts. Getting IAS and DoubleVerify on board from the start is the right move if the goal is real ad revenue rather than a press-friendly experiment.

Why This Matters Beyond the Revenue

ChatGPT's free tier has always required some form of subsidy — compute isn't free, and OpenAI's costs are staggering. Subscriptions cover some users, API revenue covers developers, but advertising is the mechanism that can make a consumer AI product genuinely free at scale. Every major platform has walked this path.

The tension, of course, is trust. Users who've come to rely on ChatGPT for research, writing, and decision-making are now sharing that interface with paid placements. OpenAI's claim that ads don't influence outputs is both essential and unverifiable from the outside. That's a dynamic advertisers will want to exploit and that users will reasonably be skeptical of, regardless of what the terms say.

For now, the test is limited in scope — a small group of advertisers, US only, during image generation only. Whether it expands into text conversations or other features will depend on how users respond and how much revenue the pilot generates. OpenAI says it will share results, but hasn't given a timeline for broader rollout decisions.