On February 9, 2026, ChatGPT ads went from industry rumour to something you could screenshot. OpenAI switched on its first live test in the US that morning, and sponsored placements began appearing for millions of free users of the most scrutinised product on the internet. Some people opened the app and felt the ground move. Others opened a media plan.
That split is exactly who this guide is for. If you use ChatGPT and want to know what changed, who sees ads, and how to make them go away, the first half is yours. If you market a business and want to know how ChatGPT advertising works as a channel, what it costs, and whether it is worth testing, keep going past that. Throughout, we flag where the platform is still wet paint, because a lot of it is.
Does ChatGPT have ads now?
Yes. OpenAI began testing ads in ChatGPT in the United States on February 9, 2026, for logged-in adults on the Free and Go tiers, Go being the lowest-priced subscription. Paid plans, meaning Plus, Pro, Business, Enterprise and Education, see none, and nor does anyone under 18. Outside the US nothing has changed yet; the test is American. Every ad carries a sponsored label and sits apart from the answer itself.
Which ad you see is not random. According to OpenAI's own help documentation, placements are chosen from the topic of the conversation, your chat history, and how you have responded to ads before. You can dismiss any single ad, delete the ad data the system holds on you, or switch personalisation off altogether.
OpenAI has also drawn public lines around the test. Ads are not supposed to influence what the model tells you. Advertisers never see your conversations, only aggregate counts of views and clicks. Sensitive territory such as health and politics is meant to stay ad-free. Those commitments matter precisely because they are checkable, and plenty of journalists are now checking.
Why ChatGPT added ads
Money, at a scale that is genuinely hard to picture. ChatGPT reached 900 million weekly active users by late February 2026, and only around 50 million of them pay for a subscription. Serving everyone else costs billions in infrastructure each year, so OpenAI frames advertising as the thing that keeps free access free.
The move follows a familiar script. Search engines and social networks each ran free for years before selling attention, and reporting from Forbes and CBS News in January traced the same arc here: compute bills mounting, a vast tier of users paying nothing, and advertising as the only bridge wide enough. What is different this time is the surface. An ad inside a conversation feels more personal than an ad beside search results, which is why the opt-outs below matter so much.
How do you turn off ChatGPT ads?
In three steps, all inside the app.
- Open Settings in ChatGPT.
- Select the ads controls section.
- Pick your exit: move to any subscription tier, since every paid plan stays ad-free, or stay free and choose the option that lowers your daily message limits and removes some tools in exchange for no ads.
The free option deserves a closer look, because it is real but not costless. You are paying with capacity instead of cash. A light user may never notice the tighter limits; a heavy free-tier user might find them more irritating than the ads were. Decide based on how often you actually hit the ceiling.
Ads are only half of the settings story, and the quieter half sits in your data controls. OpenAI's privacy policy dated April 30, 2026 formalised the company receiving purchase data from advertisers, sharing user information with marketing partners, and using your data to market its own products. PPC Land reported that marketing cookies arrive switched on by default for Free and Go accounts, with the opt-out living in the data controls area of settings, under marketing privacy. Set against the launch commitments, the pattern reads consistently: the answer layer stays clean while the data layer monetises. Two reassurances survive all of this: the content of your conversations is not handed to advertisers, and paid tiers are exempt from the whole arrangement.
Inside ChatGPT Ads Manager
Through Ads Manager, OpenAI's self-serve platform at ads.openai.com, which is still in beta. You verify a real business, upload short creative, and describe the conversations where your ad belongs. If you have been searching for how to advertise on ChatGPT, the step-by-step walkthrough lives in its own guide; the short version reads like Google Ads with the keywords removed.
Onboarding asks for more than an email address. OpenAI wants your legal business name, a physical address, and an EIN or equivalent tax registration, and agency guides report that identity verification runs through a provider called Persona.
Targeting is the biggest mental adjustment. Instead of bidding on keywords, you write context hints: freeform descriptions of the topics and conversations your ad suits. A pet insurer might describe chats about new puppies and vet bills rather than listing search terms. There is no exact match and no guaranteed placement; the system decides where you fit. Geography narrows things to country, region, DMA or postal level, and since July 7 advertisers can upload first-party audience lists, raw or hashed emails and phone numbers, with a floor of roughly 25,000 matched users and 100,000 recommended.
Creative comes in two formats, Static and Product, and users see the same shape either way: business name, image, headline, description. Late July added product cards showing prices and star ratings. On character limits we will not print numbers, deliberately: the caps are short, they are enforced live inside the ad builder, and they have already shifted during the beta.
Whatever a blog tells you today, the builder decides tomorrow.
The floor under all of this keeps moving. Conversion-optimised campaigns entered beta in June for accounts running OpenAI's pixel or conversions API, then rolled out broadly around July 24, alongside geo exclusion, advanced matching, AppsFlyer and Adjust integrations, and a bulk API for larger accounts.
| Habit | In Google Ads | In ChatGPT Ads Manager |
|---|---|---|
| Targeting | Keywords with match types | Context hints: freeform descriptions of fitting conversations |
| Placement | Bid your way onto a chosen query | No guarantees, the system decides where you fit |
| Budgets | Daily caps | Averaged across 7 days, single days can overshoot |
| Formats | Dozens across Search, Display and video | Two, Static and Product, one shared layout |
| Bidding | CPC, CPM, CPA and more | CPC, CPM or cost per conversion |
How much do ChatGPT ads cost?
Budget at least $25 per day per campaign; that is the floor named in StubGroup's guide to the platform. Bidding runs on CPC, CPM or cost-per-conversion models. From its own early campaigns, StubGroup reports paying at least $2 per click, one agency's observation rather than a market benchmark.
A late-July change complicates the daily arithmetic. Budgets are now averaged across seven days rather than capped per day, so a single Tuesday can overspend as long as the week balances out. Anyone who has managed Google campaigns will recognise both the mechanic and the reconciliation headaches it brings.
Cheap to test, hard to forecast. That is the deal right now.
Zoom out and the money question becomes a genuine disagreement. eMarketer projects US ad spending inside standalone chatbots stays under $1 billion in 2026, reaching only about $5.4 billion by 2030. OpenAI's own reported forecast is $2.5 billion in 2026, building toward $100 billion globally by 2030. The figures measure different things, one is US-only and one is global, but the gap in ambition is the story. eMarketer analyst Nate Elliott, cited by PPC Land, adds the sharpest detail: over 80% of AI-related ad spending runs alongside AI content, not inside the chatbots themselves.
The 2026 forecast gap: analysts vs OpenAI
- eMarketer: US chatbot ad spend, 2026 (under $1B)
- 1$B
- OpenAI: reported global ad revenue forecast, 2026
- 2.5$B
Should your business advertise on ChatGPT yet?
With a test budget, perhaps. With serious money, not yet. The platform is months old, its rules have changed repeatedly mid-beta, and nearly all public performance data comes from a handful of agencies spending their own money. What you buy at this stage is learning, plus early access to inventory your competitors have not touched.
There is also platform risk to price in. Perplexity, the other chatbot that tried advertising, closed to new advertisers in October 2025 and pivoted subscription-first by February 2026, its CEO arguing that ads conflict with unbiased AI answers. One rival's exit does not doom OpenAI's attempt, but it shows this business model is not settled physics yet.
And paid placement is only half of the opportunity. The same conversations where your ad might appear are conversations where the model already recommends businesses organically, based on what it knows and reads about them. That side costs structure and attention rather than media spend. It is most of what we build for clients, and our guide to AI marketing agents covers how that machinery works day to day.
A paid slot gets you into the conversation. Being the answer is what keeps you there.
The parts worth watching from here
Everything above describes a beta, and betas move. Budgets went weekly in July, product cards landed the same month, and the audience floors and formats will keep shifting under advertisers' feet. We will keep this guide current as those pieces settle.
The direction, though, looks fixed. OpenAI needs ChatGPT ads to work, an audience that size is too expensive to serve unfunded, and marketers are relearning old instincts on new ground. Whether your first move is a $25-a-day experiment or making sure the model recommends you unprompted, the worst position is not having decided.
Questions people actually ask
Who sees ads in ChatGPT?
As of the February 2026 US test, ads reach logged-in adults on the Free and Go tiers. Paid plans, meaning Plus, Pro, Business, Enterprise and Education, stay ad-free, and users under 18 are excluded entirely. Every ad sits apart from the response and carries a sponsored label.
Do ChatGPT ads change the answers you get?
OpenAI says no. Ads are selected from the conversation topic, your chat history and past ad interactions, but they appear separately from the model's answer and are marked as sponsored. OpenAI also states that conversations are never shared with advertisers, who only receive aggregate views and clicks.
What is the minimum budget for ChatGPT ads?
StubGroup's guide to the platform puts the floor at $25 per day per campaign. Since late July 2026, budgets are measured as averages across seven days rather than strict daily caps, so individual days can swing. Bidding runs on CPC, CPM or cost-per-conversion models.
Can you target keywords with ChatGPT ads?
No. Targeting runs on context hints, short freeform descriptions of the conversations where your ad belongs, with no exact matching and no guaranteed placement. You can layer on geography down to postal level and, since July 2026, first-party audience lists of roughly 25,000 matched users or more.
How long does approval for ChatGPT Ads Manager take?
Agency guides report that identity verification, handled by a provider called Persona, usually clears in 1 to 3 business days. Before applying you need an OpenAI account, your legal business name, a physical address, and an EIN or equivalent tax registration.
