Separate paid placements from the organic text and citations in an AI answer. Evaluate the provider’s current advertising terms for any paid placement, and record it separately from sampled brand mentions. An ad beside an AI answer is not an organic recommendation.
- OpenAI says ChatGPT ads testing begins in the US on February 9, 2026 for Free and Go users, and states plainly that ads do not influence ChatGPT's answers.
- Perplexity tested sponsored follow-up questions in 2024, then abandoned advertising entirely in February 2026, saying users need to believe they are getting the best possible answer.
- Google makes existing Search, Shopping, Performance Max, and App campaign ads eligible above or below AI Overviews, with within-overview placements available when relevant.
- None of these products let you pay to be the brand the model names in its organic answer. There is no cost-per-mention and no budget dial for recommendations.
- 75% of Americans say they would trust AI shopping less if recommendations were swayed by brand dollars (Quad / The Harris Poll, Feb 2026), which is exactly why platforms wall ads off from answers.
What you can and cannot pay for in AI search
The confusion is reasonable, because two different things now live on the same screen. The first is a paid ad unit: a labeled, sponsored placement that sits near an AI answer, sold through an auction, billed per click or impression. The second is the organic recommendation, the brands the model actually names when it writes "the best options are." You can buy the first. You cannot buy the second. And even inside that organic answer, being named is not the same as being the pick (the gap between AI mention and recommendation). Parse tracks AI visibility across ChatGPT and Google AI Mode, analyzing 3.85 million AI responses and more than 47 million citation observations across 577K+ brands, and none of those in-answer recommendations carry a price tag. There is no rate card for being mentioned, no targeting console for citations, no way to outbid a competitor into the body of a response. When someone asks whether they can pay to appear in AI search, the honest answer is: you can pay to appear beside it.
How ChatGPT ads actually work
OpenAI says ChatGPT ads testing begins in the US on February 9, 2026 for users on the Free and Go plans. Its Help Center says ads can appear below a response, visually separated from the answer and labeled as sponsored. Users on Plus, Pro, Business, Enterprise, and Education plans do not see them, under-18 users are excluded, and sensitive topics like health and politics are kept ad-free. OpenAI also lists Canada, the United Kingdom, Australia, and New Zealand among the countries where ads may appear as the test expands. The principle OpenAI repeats matters more than the placement: ads do not influence ChatGPT's answers, and advertisers cannot pay to shape, rank, or alter the response. That is also why buying a ChatGPT ad is not the same as getting recommended inside ChatGPT shopping, which pulls from organic signals, not your ad spend.
Why Perplexity walked away from advertising
Perplexity was first to try the model and first to abandon it. Starting in 2024 it ran two ad formats: sponsored follow-up questions inside the "Related Questions" section, where roughly 40% of platform queries land, and sponsored video alongside answers. Early partners included Indeed, Whole Foods Market, and Universal. Then, as reported by the Financial Times in February 2026, the company phased the program out entirely and said it has no plans to bring it back. The reasoning is the cleanest statement of the trust problem anyone has offered: a user needs to believe this is the best possible answer, and once ads appear, people start doubting everything. Perplexity reframed itself around being "in the accuracy business" and now leans on subscriptions and shopping features that take no transaction cut. For brands, the lesson is blunt: even the platform most willing to experiment with paid placement concluded it threatened the product.
How Google sells ads around AI Overviews
Google is the outlier, because it has the most to monetize and the deepest ad machinery to do it. AI Overviews now appear across a large share of Google searches, and Google says existing text, Shopping, local, and app ads in Search, Shopping, Performance Max, and App campaigns are eligible to show above or below AI Overviews in markets where AI Overviews are available. Google also supports ads within AI Overviews when the placement is relevant to both the query and the answer. Everything is labeled sponsored. The important detail is the same as everywhere else: these are ad placements contextually integrated around the AI answer, not a lever that edits which brands the generated summary names organically.
If you want to see how AI engines describe your own brand, run a free brand check. It takes a minute.
What the paid layer looks like across platforms
The three platforms differ on how aggressively they monetize, but converge on one rule. The table below sets the products side by side; read the last column first, because it is the one that answers the headline.
| Platform | Paid product (2026) | Where it appears | Can it change the organic recommendation? |
|---|---|---|---|
| ChatGPT | Sponsored ads (Free and Go tiers) | Below the response, visually separated, labeled | No. Ads run on separate systems from the model. |
| Google AI Overviews / AI Mode | Search, Shopping, Performance Max, App, and related AI campaign inventory | Above, below, and, when relevant, inside AI Overviews | No. Sponsored units sit around the generated answer. |
| Perplexity | None (program ended Feb 2026) | Previously sponsored follow-up questions and video | No, and there is now no ad product at all. |
Across every surface, the recommendation written into the answer is off the auction. The paid layer buys attention adjacent to the answer; it does not buy authorship of it. That single column is the difference between an ad strategy and an AI visibility strategy.
The trust ceiling that limits paid AI placements
Platforms wall ads off from answers because their own users demand it. A February 2026 survey of 2,180 US adults by Quad and The Harris Poll found 75% would trust AI shopping less if recommendations were swayed by brand dollars, and the same share would trust the brands themselves less for paying to influence an AI agent. That is not a soft preference; it is the constraint that shapes every platform's product decisions, from OpenAI saying ads do not influence ChatGPT's answers to Perplexity exiting ads outright. The commercial logic follows the trust logic: an answer engine that can be bought is worth less than one that cannot.
The reason you cannot pay to be recommended is not that platforms haven't figured out how to charge you. It is that the recommendation only has value because it cannot be bought. The moment a model's answer is for sale, the answer stops being worth reading, and the platform loses the asset it monetizes everywhere else.
Why the in-answer recommendation stays earned, not bought
The brands a model names come from a different supply chain than its ad inventory. When ChatGPT or an AI Overview writes "consider these options," it is assembling that list from the entities, citations, and earned mentions it encountered while retrieving sources, not from a billing system (the source domains those answers are built on). The model needs a clear record of who you are, third-party coverage on the sources it trusts, and content it can extract cleanly. How ChatGPT decides which brands to recommend walks through the retrieval mechanics, and which domains AI models cite most shows where those mentions actually come from. The uncomfortable part for budget holders is that the most influential signal is earned media, which you cannot purchase as media: earned media for AI citation covers how to pursue it deliberately. Ads can sit next to that answer. They cannot write it.
How to split a budget between ads and earned AI visibility
Treat the two layers as two line items with different jobs. AI search ads are a paid-media decision: test them like any new channel, measure incremental clicks and conversions against cost, and expect the labeled, separated placement to behave more like display than like a recommendation. They are worth piloting where your category shows commercial intent and your competitors are bidding. But the budget that compounds goes to the earned layer, because that is what determines whether the model names you at all. Fund the entity work, the third-party coverage, and the content structure that gets you cited, then measure the result so you know it is moving. AI citation gap analysis shows which sources feed competitors but not you, and share of model turns "are we even in the answer" into a number you can track over time. Spend on ads for the slot you can rent; invest in citations for the slot you have to earn.
Of Americans say they would trust AI shopping less if recommendations were swayed by brand dollars.
Existing Search, Shopping, Performance Max, and App campaign ads can appear around AI Overviews when relevant.
Where eligible Search, Shopping, Performance Max, and App ads can appear above or below AI Overviews.
Sponsored answers remaining on Perplexity after it abandoned advertising in February 2026.
See where you actually stand in AI answers
If you cannot buy the recommendation, the first move is to measure whether you are in it. Pull your category's commercial prompts, see how often the models name you versus the competitors you lose to, and find the sources feeding them that are not yet feeding you. Search your brand in Parse to see where you appear across ChatGPT and Google AI Mode, and which earned signals, not ad budgets, are deciding the answer.