ChatGPT ads get a visual format as OpenAI widens measurement

OpenAI will begin testing a new visual advertising format inside ChatGPT later this month in the United States, the company said on 5 October, and it is expanding the measurement tools advertisers can use to judge whether those ads do anything.

The ads will use images to show product inspiration, product usage or the experiences a product makes possible, appearing during image generation in ChatGPT. OpenAI says the ads will be clearly labelled and kept separate from the image a user is creating, and that advertising does not influence the answers ChatGPT gives. Testing starts later this month with an initial group of advertisers.

Why advertising matters to OpenAI

OpenAI frames the move around reach. It says ChatGPT now serves 1.2 billion people each week, making it the largest AI-native consumer platform, and argues that advertising lets it serve a much larger market than subscriptions alone can. That is the same pitch every large platform makes, and it lands differently here: OpenAI has spent years telling regulators and users that the assistant is not an ad business.

Measuring what the ads are worth

The second half of the announcement is about proof. OpenAI is adding integrations with Hightouch, Tealium and LiveRamp so advertisers can push conversion data from the systems they already run into ChatGPT Ads, and it now supports attribution partners across web and app, including AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge and Tenjin. Full-funnel and advanced measurement partners include Fospha, Measured and INCRMNTAL.

Incrementality work, which tries to separate sales an ad actually caused from sales that would have happened anyway, is described as still in progress. That wording matters more than the partner list. Advertisers will not move serious budgets onto an unproven format just because it is new; they will move it once someone can show the lift.

Our opinion

The interesting question is not whether ChatGPT can carry ads, but what happens to the product once it does. OpenAI has built its reputation on an assistant that answers rather than sells, and it is now threading a commercial layer through the same surface. The promise that advertising will not influence answers is the right one, but it is also precisely the promise that will be hardest to audit from the outside, and the measurement partnerships announced here measure advertiser outcomes rather than user trust.

There is a reasonable version of this. Ads in a free tier are a fair trade for people who would otherwise pay or go without, and clearly labelled image ads during image generation are a far less intrusive idea than sponsored answers. The trouble is scope. Once the plumbing exists, the pressure to widen it is constant, and the distance between a clearly labelled ad in the corner of a canvas and a nudge inside a recommendation is much shorter than this announcement implies.

Photo by cottonbro studio on Pexels.