They sell ads in ChatGPT now
For most of the last two years the honest sales pitch for AI visibility included a line that consultants enjoyed saying: you cannot buy your way into an AI recommendation. There is no ad slot. It has to be earned.
That line needs revising, and I have revised it on this site rather than waiting to be corrected by a client.
What actually happened
OpenAI began showing ads in ChatGPT in February 2026, to free-tier users in a handful of countries. In May it opened a self-serve Ads Manager, dropping the pilot's minimum spend entirely. Advertisers now build campaigns with budgets, bids, pacing and conversion tracking, in a three-level campaign structure any Google Ads buyer would recognise in about four minutes.
Since then: integrations with HubSpot as first CRM partner and Shopify as first ecommerce partner, an adtech partnership with Criteo, and testing of Sponsored Agents — where clicking an ad opens a conversation with a business-sponsored agent rather than a landing page.
This is not an experiment any more. It is an advertising platform.
The detail everyone is skipping
OpenAI is specific about where the ads sit. They appear in clearly labelled units below the response, and they do not influence the answer itself.
The recommendation and the advertisement are two different things on the same screen.
If you ask ChatGPT for a dentist in your city, the practices it names are still chosen the way they were last year: from sources that corroborate each other, with consistent facts, structured so a model can read them without guessing. Underneath that answer there may now be a labelled ad. The advertiser paid for the slot. They did not pay to be in the paragraph above it.
So the accurate version of the old line is narrower and, I think, more interesting: you can buy an advertisement, you cannot buy the recommendation.
Why this makes earned placement worth more
The instinct is that ads devalue organic presence. Search taught the opposite lesson over twenty years.
When a results page carries advertising, people learn to identify it and discount it. The label that protects the user — Sponsored — is also the label that tells them this one paid to be here and the others did not. Trust concentrates on the part that could not be bought.
An AI answer is a sharper version of the same thing, because there are only three or four names in it rather than ten blue links. Being one of those names, above a labelled advertisement, is a stronger position than being the advertisement. It always was in search. It is more so here.
What it does change
Three things, and they are real.
Your competitors now have a way to appear without earning it. Not in the answer, but on the screen, beneath the answer, at the moment your customer is deciding. If you are not in the answer and they are in the ad slot, you are absent from that screen entirely.
The cost of doing nothing went up. Previously, invisibility in AI answers meant being absent. Now it can mean being absent while a competitor is present and paid for it.
And a measurement problem arrived. If traffic from ChatGPT increases, you now need to know whether it came from the answer or from an ad, because those cost very different amounts. Anyone reporting "AI visibility" as one number from here on is measuring two things and calling them one.
What I would do about it
If you are already earning citations, nothing urgent. Keep going. Your position just became more valuable relative to a paid one.
If you are invisible in AI answers, the arrival of ads is not a shortcut. Buying the slot beneath an answer that names your competitor is an expensive way to look like the alternative. Fix the answer first.
And if you run paid media already, this is worth a test budget rather than a strategy. It is a new surface with genuine intent behind it and thin competition, which is the familiar early-platform trade: cheap attention, immature measurement.
Common questions
Can you pay to appear in a ChatGPT answer?
You can pay for an advertisement, not for the recommendation. OpenAI shows ads as clearly labelled units below the response and states they do not influence the answer itself. Which businesses get named in the answer is still determined by whether a model finds consistent, corroborated facts about them across independent sources.
Do ChatGPT ads make AI visibility work less valuable?
The opposite. When a results surface carries advertising, users learn to identify and discount it, and trust concentrates on the part that could not be bought. With only three or four businesses named in a typical AI answer, being one of them above a labelled advertisement is a stronger position than being the advertisement.
Should my business advertise in ChatGPT?
If you are already named in AI answers, it is worth a test budget rather than a strategy — a new surface with real intent and thin competition, but immature measurement. If you are invisible in AI answers, fix that first. Buying the slot beneath an answer that names your competitor is an expensive way to look like the alternative.
Find out which side you are on. The free audit asks a live assistant real customer questions and shows whether your business comes back — and which sources it used to decide.
Run the free auditA note on changing your own copy
I had "there is no ad slot in an AI recommendation" written in two places on this site. It was true when I wrote it and it is not true now, so it is gone.
That is worth saying out loud because this field moves faster than the confident claims made about it, and a consultant whose website still argues something the market disproved three months ago is telling you something about how closely they are watching. Check the people advising you on this. The line they used last quarter should not still be there.