Yes, on Google, and more than most people realise. Partly on ChatGPT. And in every case what you are buying is a labelled advertisement, not a citation, which are two different pieces of inventory on the same screen.
That distinction is the whole answer, and it matters because the budget lines look interchangeable in a plan and are not interchangeable in effect. All the facts below were checked against each vendor's own live documentation on 19 August 2026.
Key Takeaways
- Google Ads documents ads showing above, below and within AI Overviews, and states you cannot opt out of serving them there.
- Ads inside the Overview are documented for English on mobile and desktop in twelve named countries. Above and below runs everywhere AI Overviews do.
- Google says it does not currently offer segmented reporting for ads shown within AI Overviews, and reports them as Top Ads. You are paying for a placement you cannot isolate in your own dashboard.
- OpenAI documents a crawler named OAI-AdsBot whose job is validating pages submitted as advertisements on ChatGPT.
- Perplexity and Microsoft Copilot could not be confirmed from their own pages today, and this post says so rather than filling the gap from secondary reporting.
What Google actually sells
Google's help page on ads and AI Overviews is unusually specific, and worth reading rather than summarising, because most coverage of it stops at "Google put ads in AI Overviews" and misses the operational details.
| Documented | |
|---|---|
| Placement | Above, below, and within the AI Overview |
| Above and below | All markets where AI Overviews are available |
| Within the Overview | English, mobile and desktop, in twelve named countries including the US, Canada, Australia and India |
| Formats | Text, shopping, local and app ads from existing Search, Shopping, Performance Max and App campaigns |
| Excluded | Sensitive verticals including adult, alcohol, gambling, finance, healthcare and politics |
| Opt out | Not available |
| Reporting | No segmented reporting for ads within AI Overviews. Reported as Top Ads |
Two rows deserve more attention than they get.
The opt out row means this is not a campaign type you switch on. If you are running eligible Search, Shopping or Performance Max campaigns in a market where this is live, your ads are already eligible for these placements. There was no decision to make and there is no decision to unmake.
The reporting row means you cannot tell how much of your spend landed there. Google's page says it does not currently offer segmented reporting when ads show within Search AI Overviews, and that they are reported as Top Ads. The clicks are in your account. Which surface produced them is not.
What OpenAI sells, and the crawler that gives it away
OpenAI does not publish an ads help centre with the detail Google's has. It does publish a crawler list, and that list answers the question sideways.
OpenAI's crawler documentation names four agents. Three are the ones people argue about: GPTBot for training, OAI-SearchBot for ChatGPT's search feature, ChatGPT-User for fetches a person triggers. The fourth is OAI-AdsBot, and OpenAI describes its job as validating the safety of web pages submitted as advertisements on ChatGPT. The page adds that what it collects is not used to train foundation models.
A crawler that validates submitted advertisement landing pages only exists if there are submitted advertisement landing pages. That is a paid placement pipeline described in the plainest possible terms, in a document written for webmasters rather than for buyers.
What it is not is a route into an organic answer. The agent that feeds ChatGPT's search answers is a different one with a different name, and OpenAI keeps them separate in its own documentation.
What could not be confirmed today
Two of the engines worth asking about did not yield a citable answer on 19 August 2026, and this section exists rather than being quietly dropped.
Perplexity publishes a blog post on its own site about experimenting with advertising, and the page refused an automated request. Secondary coverage describes sponsored follow-up questions. That is probably right and it is not something to state as fact when the primary source could not be opened.
Microsoft Copilot was worse in an interesting way. The Microsoft Advertising help page for ads in Copilot returned a shell containing navigation, a copyright line and cookie preferences, with the substance loaded afterwards by script. Anything reading that page without executing JavaScript, which includes plenty of the agents this blog is about, gets nothing.
Neither gap changes the answer. Both are named so a reader knows exactly where the evidence stops.
The paid slot is not the cited slot
The reason to keep these apart is not tidiness, it is that they respond to completely different work.
A paid placement is bought, labelled, and appears because your bid won an auction. A citation appears because a retrieval step found a page and a ranking step chose it, and no vendor publishes a mechanism by which ad spend influences that choice. Both vendors describe the two as separate systems, and there is nothing in either documentation set that connects them.
The practical consequence is that the two answer different questions. Buying the ad answers "can I be on this screen". It says nothing about whether the answer above it names you, or names four competitors instead, which is the question your customer's eye actually lands on.
That is also why ad reporting cannot substitute for visibility measurement, even where the reporting is good. An impression means your ad was shown. It does not tell you what the answer said.
The measurement gap this leaves
Put the two documented facts side by side and the shape of the problem is clear.
Google will place your ad inside an AI Overview, will not let you opt out, and does not break out the placement in reporting. So the surface with the most attention on the page is one you are already paying for and cannot isolate.
Meanwhile the organic half of that same Overview, the part that says who the answer trusts, is not in any advertising dashboard at all. It is not in Search Console either. The only way to know what an answer says about you is to ask the question and read the answer.
That is a measurement job rather than a media buying job, and it is the one nobody's billing system does for you.
What to do this week
- If you run Search, Shopping or Performance Max in a market where AI Overviews are live, accept that some of your spend is already serving there and that you cannot separate it. Plan on total performance rather than on a placement line you will not get.
- Do not shift budget on the theory that ads improve organic citation. Nothing either vendor publishes supports it, and Google's own documentation describes the two placements as distinct.
- Check whether your ad landing pages are reachable by the crawlers that validate them. If a robots.txt rule blocks the agents named in OpenAI's crawler list, find out before a campaign does.
- Measure the organic answer separately, because no advertising report contains it. Run the free check on your domain and the topic you want to be known for. No account and no card. It puts a real buyer question to Google's AI Overview and to ChatGPT and tells you whether the answer named you, and which sources it used instead.