The answers now sell ad space. The citations don’t.

The timeline moved fast. OpenAI began showing ads to US free-tier users on 9 February 2026 — clearly labelled boxes beneath answers, matched to the conversation’s context. A self-serve Ads Manager followed on 5 May with the pilot’s $200,000 minimum removed; reported entry pricing sits around $3–5 per click. And on 11 August OpenAI announced expansion to the UK, Mexico, Brazil, Japan, and South Korea, with Canada, Australia, and New Zealand piloting. Google, meanwhile, launched shopping ads with Direct Offers inside AI Mode in February. The era of the ad-free AI answer lasted roughly three years.

A channel worth testing — carefully

Conversational intent is deeper than keyword intent: a person mid-conversation about their problem is closer to a decision than a person typing two words into a search box. Early access is uncrowded, pricing is not yet bid up, and measurement exists (a first-party conversion pixel). For brands whose buyers demonstrably ask assistants for recommendations, a small, measured test belongs on the 2026 roadmap. It is also honest to note the counter-evidence: Perplexity paused its own ad programme after returns disappointed, and OpenAI showed pilot ads to fewer than a fifth of eligible users on a given day — the channel is real but young, and nobody should move core budget yet.

The line OpenAI drew — and what it means

OpenAI’s documentation is explicit: ads do not influence ChatGPT’s answers; the ad system runs separately from the model. Google labels its AI Mode placements as sponsored. Which means the answer itself — the brands the assistant names, the sources it cites — remains an editorial space no budget can enter. Two slots now exist side by side: a paid one, available to anyone with money, and an organic one, available only to brands the model trusts. Users learned the difference on Google over twenty years; they will learn it faster here.

The strategic read

Advertising arriving in the answers does not weaken the case for GEO — it completes it. The paid slot creates a floor of visibility you can rent; the cited slot carries the trust, and its scarcity now has a market price next to it for comparison. The playbook is the one that worked in search: rent reach while you build the organic asset, and let the asset compound while competitors keep paying rent. Our FIB AI Visibility Check tells you which slot you currently occupy — if either.

How FIB approaches it

We treat AI ads as a test channel and AI citations as the asset — and we measure both. If your category’s buyers are already asking assistants what to choose, get in touch.

Sources

OpenAI, Testing ads in ChatGPT, updated 11 August 2026 · OpenAI Ads Manager launch coverage, May–July 2026 (Segwise, Jolly Good Web, Choice OMG) · Digital Applied on Google AI Mode Direct Offers, April 2026 · eMarketer on Perplexity’s paused ads and Gemini, 2026.