On 18 August 2026, OpenAI announced that ChatGPT Ads would be available on 24 August across 31 European markets, Germany, France, Spain, Italy, Sweden, Norway, Denmark, the Netherlands and Austria among them. Six months after testing began in the United States, ChatGPT's ad platform is landing in Europe — and not quite in the shape most coverage implies.
Two details change a great deal, and both went largely unnoticed. First: at launch, a European advertiser cannot build a campaign on their own. Second: personalised ads are not available in Europe at first. Both come straight from OpenAI, and together they determine what a business can actually do on 24 August and what a user will actually see in their feed.
Everything below comes from OpenAI's own pages: the 18 August announcement, the pilot page updated on 11 August, the help centre, and the ad policies. Every claim points to its source. Where OpenAI says nothing, I say so rather than filling the gap.
📌 TL;DR
ChatGPT Ads opens across Europe on 24 August 2026. At launch, access runs through OpenAI's sales team, a partner agency or a technology partner; self-serve arrives "later this quarter". Buying is CPM or CPC, with a recommended starting bid of 3 to 5 US dollars per click. Whole industries are barred at launch. And on the user side, only the Free and Go plans see ads — with a free way to switch them off, in exchange for lower limits.
In this article
- What exactly happens on 24 August
- How a European advertiser gets in
- What a ChatGPT ad looks like
- What it costs: CPM, CPC and the starting bid
- Targeting is not keywords — and Europe has one rule fewer
- What you are not allowed to promote
- Measuring what it returns
- User side: who sees ads, and how to switch them off
- What we still do not know
- FAQ - Frequently asked questions
What exactly happens on 24 August
The announcement itself is one sentence: "On August 24, ChatGPT Ads will be available in 31 European countries." Nine are named — Germany, France, Spain, Italy, Sweden, Norway, Denmark, the Netherlands and Austria. The other twenty-two are not, and OpenAI has not published the full list.
Read the date for what it is: this is the date the ad platform opens to advertisers, not the date every European user starts seeing ads. OpenAI commits to no user-facing rollout schedule at all. That distinction matters if what you actually want to know is when your own ChatGPT will start looking different.
To place this launch on the product's trajectory, here is the official timeline as OpenAI has documented it, update by update, on its pilot page.
| Date | Milestone |
|---|---|
| 9 February 2026 | Testing begins in the United States, on the Free and Go plans, adult users only |
| 26 March 2026 | Expansion announced to Canada, Australia and New Zealand |
| 7 May 2026 | Expansion announced to the UK, Mexico, Brazil, Japan and South Korea |
| 11 August 2026 | Those five markets go live |
| 24 August 2026 | Opening across 31 European markets |
OpenAI also says that "tens of thousands of advertisers" have used the platform to date, without a precise figure or a breakdown by market. The brands showcased on its advertiser site are American: Newegg, Best Buy, Lowe's and VistaPrint.
How a European advertiser gets in
This is the point most coverage skips, and it is the most concrete one. Here is OpenAI's exact wording: "Initially, ChatGPT Ads will be accessible through OpenAI's Ads Solutions team, partner ad agencies, and technology partners. Self-serve through Ads Manager will launch later this quarter."
In other words: on 24 August, a European small business does not log into an interface and launch a campaign that afternoon. It goes through a salesperson, an agency, or a technology partner.
That is not an interpretation. OpenAI's Ads Manager availability page, updated roughly a week before this article was published, lists nine countries where the ad manager is available: Australia, Brazil, Canada, Japan, South Korea, Mexico, New Zealand, the United Kingdom and the United States. No European Union country appears on it.
Two different things share a name
"ChatGPT Ads available in Europe on 24 August" and "Ads Manager available in Europe" do not mean the same thing. The first is announced; the second is not, yet. If you see an article promising self-serve access on the 24th, check the official availability page: it is the authority here, and it carries a date.
In practice, two routes remain open to a European business that wants to move now. The first is the advertiser interest form on ads.openai.com, which asks for your headquarters country, the countries you want to advertise in, and your organisation type — agency, advertiser, technology partner or other. The second is to work through an agency that is already a partner.
When self-serve does open, the onboarding OpenAI describes is conventional: sign in with an OpenAI account tied to a work email address or create one, add your brand logo in Settings, set up a billing profile and payment method, complete account verification, then invite team members. Nothing exotic compared with Google Ads or Meta.
What a ChatGPT ad looks like
The help centre describes the object precisely. A ChatGPT ad appears below a ChatGPT response, never inside it, and carries six elements:
- the advertiser name;
- the favicon, meaning the advertiser's logo;
- a headline;
- body text, the description;
- a landing page;
- an image asset, the creative itself.
You may see one or several ads under a single response, and that unit can group items from one advertiser or from several. OpenAI adds that in long conversations the system also weighs overall context and user experience — without spelling out what that means for volume.
Two exclusions are worth noting, because they are not obvious. During this test, ads do not appear in the ChatGPT Atlas browser, and Temporary Chats show no ads either. If you mostly use ChatGPT in temporary conversations, your experience will not change.
Finally, the ad policies explicitly ban interface mimicry: an ad cannot imitate the look, functionality or voice of ChatGPT in a way that could reasonably lead users to believe it is part of the product. That is the counterpart to OpenAI's promise that ads stay "clearly labeled and distinct from ChatGPT's responses".
What it costs: CPM, CPC and the starting bid
On pricing, OpenAI is more specific than you might expect from a platform still in beta. Buying is on a cost per thousand impressions or cost per click basis, depending on the objective you pick.
| Objective | Optimises for | Billing |
|---|---|---|
| Reach | Impressions | Every 1,000 impressions (CPM) |
| Clicks | Clicks | Each valid click (CPC) |
The maximum bid is set at ad set level: a max CPM for Reach campaigns, a max CPC for Clicks campaigns. For CPC campaigns, OpenAI recommends starting with a maximum bid of 3 to 5 US dollars per click. That is the only published cost benchmark the platform offers, and it comes straight from its documentation rather than from an agency estimate.
Selection runs through relevance-weighted second-price auctions. The mechanism will be familiar to anyone who has run search: the highest bid does not mechanically win, the bid × relevance combination does. Ads Manager also flags whether a bid is likely to be too low to compete.
One vocabulary note, to avoid a confusion. The 18 August announcement mentions a third mode, optimised cost per click (oCPC), presented as the most recent addition after CPM and then CPC. The help centre page describing the basics still lists only the Reach and Clicks objectives. I found no official document stating whether oCPC will be available at the European opening: treat it as something to check when configuring a campaign, not as a given.
On minimum budget, however, nothing. No public OpenAI page sets an entry threshold. If an article quotes you one, ask it for its source.
Targeting is not keywords — and Europe has one rule fewer
This is probably the most structural difference from Google Ads, and the one most likely to trip up people arriving from search. In ChatGPT Ads, you do not target keywords.
At ad group level, the advertiser supplies context hints: descriptions of the conversations, topics or keywords their products may be relevant to. The help centre is explicit about their status: these hints "help guide ad matching, but they are not exact-match keywords and do not guarantee delivery in specific conversations".
The system then combines several signals: the context and intent of the current conversation, the landing page, the ad headline and text, the hints the advertiser supplied, and — when ad personalisation is on — some signals from broader ChatGPT usage. The 18 August announcement adds that geographic targeting and custom audiences were introduced over the past months.
The point that really sets Europe apart
OpenAI's help centre states that "personalized ads are not initially available in the European Economic Area (EEA) or Switzerland". In practice, an ad seen from Europe draws on the context of the current conversation, and not on chat history, memory or ad history. OpenAI does not give a reason for this limitation — so I will not supply one on its behalf, even though the European regulatory calendar invites the hypothesis.
For an advertiser, that means performance in Europe will depend far more, at the start, on the quality of the context hints and the clarity of the offer than on any audience refinement. For a user, it means an ad seen in Europe reveals considerably less about you than one seen in the United States.
On the European framework around all of this — transparency obligations, the timeline, what already applies to businesses — our article on the AI Act's concrete obligations for SMEs separates what is enforceable today from what is not yet.
What you are not allowed to promote
OpenAI's ad policies, updated on 10 August 2026, are restrictive — noticeably more so than a mature ad platform would lead you to expect. During the initial test period, advertising is "primarily limited to consumer categories".
| Status | Sectors |
|---|---|
| Allowed at launch | Lifestyle and household goods, local services, travel and experiences, digital products or education |
| Case by case | Financial services, health and medicine, legal services — manual approval, phased rollout |
| Prohibited at launch | Dating and sexual content, health claims, alcohol and drugs, gambling, political content, and every category not listed above |
Translated for a European business: a holiday rental, a tradesperson, a travel agency, a software vendor or a training provider fall inside the permitted scope. An insurance broker, a law firm or a clinic go through case-by-case manual approval. A betting site, a wine merchant or a dating app do not get in at all.
Two cross-cutting rules deserve attention, because they get otherwise-compliant campaigns rejected. First, the standards apply to every component of the ad, landing page included: OpenAI's own example is a meal-delivery ad pointing to an alcohol-delivery service. Second, ads must be truthful, particularly about capabilities, pricing, results and comparisons with competitors.
On placement, OpenAI does not run ads next to conversations about personal health, mental health or politics, nor across a range of contexts it describes as user-sensitive or brand-unsafe. Political advertising is not currently permitted on ChatGPT at all.
Measuring what it returns
On measurement, the platform has closed part of its gap over recent months. The 18 August announcement names three pieces: OpenAI Pixel, a Conversions API, and third-party measurement integrations.
Inside Ads Manager, available reporting covers impressions, clicks, spend, click-through rate, average CPC, average CPM, and conversions where conversion measurement is configured. That data appears in a table by campaign, ad group or ad, in trend charts, and exports to CSV with daily or cumulative values over a chosen date range.
One practical detail that keeps you from flying blind: static tracking parameters added to landing page URLs, such as UTM parameters, are preserved through ad clicks. So you can isolate ChatGPT Ads traffic in your existing analytics tool without depending solely on OpenAI's reporting. For a small business wanting to test the platform without committing to a heavy integration, that is the simplest way to know what actually lands on the site.
As for what gets passed on, OpenAI is categorical and repeats it in several places: advertisers receive only aggregated, non-identifying reporting. They have no access to conversations, chat history, memories, your name, email address, precise location, IP address or sensitive information.
User side: who sees ads, and how to switch them off
If you are reading this as a ChatGPT user rather than an advertiser, this is your section. It comes down to three facts.
1. Not everyone is affected
Ads may appear for users on the Free and Go plans. The Plus, Pro, Business, Enterprise and Edu plans do not include advertising. No ads are served to accounts identified as belonging to people under 18, based on account-level age information and, where available, age prediction.
2. There is a free ad-free option
This is the least known part, and it costs nothing. On the free plan, you can switch to an ad-free experience. The trade-off is stated plainly: lower usage limits — fewer messages per day — and reduced access to some features, with OpenAI citing image generation and deep research.
The path, as the help centre describes it:
- open Settings in ChatGPT;
- select Ad controls;
- choose to switch plan to go ad-free;
- choose "Lower message limits";
- confirm the change.
Reverting is built in: from the usage-limit banner, or from the same settings. Note, if you are on the Go plan: the ad-free option belongs to the free plan. You would need to move back to Free to use it, or upgrade to Plus or Pro, which are ad-free by design.
3. Turning off personalisation is not the same as removing ads
The two settings get confused constantly, and OpenAI takes care to separate them. Turning off personalisation changes how ads are selected; the ad-free option removes them. If you switch off personalisation, you keep seeing ads, but chosen on the context of the current conversation, without drawing on your other conversations, your ad history or your topics.
A useful reminder for Europe: according to the help centre, personalised ads are not available in the EEA and Switzerland at first anyway. So this setting has less effect here than elsewhere at the start.
Finally, the three-dot menu on an ad lets you hide it, report it, find out why it is being shown, or share it with ChatGPT to ask questions about it — by default, ChatGPT cannot see the ads presented to you. And if you clear your ad data, OpenAI states it immediately stops being used and is removed from its servers within 30 days.
For the wider background — why OpenAI is introducing advertising, how ChatGPT's business model is shifting — our January article on ads in ChatGPT still holds on the principles, but it was written before the European opening: the dated material on this page complements it.
What we still do not know
Worth saying plainly: this subject has more grey areas than the headlines suggest. Here is what I could not verify against an official source, and which you will therefore not find asserted above.
- The full list of the 31 European markets. OpenAI names nine. The rest are not published to date.
- The exact self-serve opening date in Europe. "Later this quarter" is the only indication given.
- The minimum budget. No entry threshold is published.
- Observed CPM or CPC levels in Europe. The platform has not served there yet; any figure circulating today is a projection, not a measurement.
- ChatGPT user numbers per European country. I found no figure published by OpenAI. Those in circulation come from blog posts or social media, without a primary source.
- Why personalised ads are excluded in the EEA and Switzerland. OpenAI states the fact without explaining it.
- Whether oCPC is available at the European opening. Mentioned in the announcement, absent from the page describing campaign objectives.
This list will move quickly, which is normal for a beta product. The useful habit, if you need to decide something on the basis of any of it: open the relevant help centre page and look at its update date, which is displayed.
FAQ - Frequently asked questions
The questions that have come up most since the 18 August announcement, with short, sourced answers.
When do ads start appearing in ChatGPT in Europe?
OpenAI announced on 18 August 2026 that ChatGPT Ads would be available on 24 August 2026 across 31 European markets. That announcement covers opening the ad platform to advertisers: OpenAI gives no date for when any individual European user will see their first ad.
Will I see ads if I pay for ChatGPT Plus?
No. OpenAI states that the Plus, Pro, Business, Enterprise and Edu plans do not include advertising. Ads may appear for users on the Free and Go plans, and never on accounts identified as belonging to people under 18.
Can I remove ads without paying for a subscription?
Yes. OpenAI offers an ad-free option on the free plan in exchange for lower usage limits: fewer messages per day and reduced access to some tools, such as image generation or deep research. The setting sits under Settings, then Ad controls.
Are ChatGPT ads personalised in Europe?
Not initially. OpenAI's help centre states that personalised ads are not available at first in the European Economic Area or Switzerland. In practice, an ad seen from Europe draws on the context of the current conversation, not on chat history or memory. OpenAI does not give a reason for this limitation.
Can a European advertiser build a campaign themselves on 24 August?
Not at launch. OpenAI states that ChatGPT Ads will first be accessible through its Ads Solutions team, partner ad agencies and technology partners, and that self-serve access via Ads Manager will launch later in the quarter. As of 20 August 2026, the Ads Manager availability page lists nine countries, none of them in the European Union.
How much does a ChatGPT ad cost?
Buying is on a cost per thousand impressions or cost per click basis, with a maximum bid set at ad set level. For cost per click campaigns, OpenAI recommends starting with a maximum bid of 3 to 5 US dollars per click. Selection runs through relevance-weighted second-price auctions. OpenAI publishes no minimum budget.
Do advertisers see my ChatGPT conversations?
No. OpenAI states that advertisers have no access to conversations, chat history, memories, your name, email address, precise location, IP address or sensitive information. They receive only aggregated, non-identifying reporting, such as total views or clicks.
Which industries are not allowed to advertise on ChatGPT?
OpenAI's ad policies limit the launch to consumer sectors: lifestyle and household goods, local services, travel and experiences, and digital products or education. Financial services, health and medicine, and legal services may be approved case by case. Every other category is prohibited at launch, including dating, alcohol, gambling and political content.
What to take away
24 August 2026 is a commercial opening date, not an interface switch. If you are an advertiser in Europe, the question to settle this week is not "how do I launch my campaign" but "do I go through the interest form, through a partner agency, or do I wait for self-serve". If you are a user, it is simpler still: on Free or Go you will see ads and you can switch them off for free; on Plus, Pro, Business, Enterprise or Edu, nothing happens.
What makes this launch worth following is precisely what is missing: no personalisation in Europe at the start, whole sectors closed, no self-serve, no observed cost benchmark. A platform in that state gets judged over three months, not three days.
Check for yourself
Everything in this article comes from public OpenAI pages, and they are dated: the European rollout announcement, the pilot page and its update log, the ChatGPT Ads basics on the advertiser side, the user FAQ, the Ads Manager availability page and the ad policies. Each one shows its last update date: on a beta product, that is the most useful information on the page.
Want to follow how this develops? Subscribe to the EverydayAITech newsletter for weekly analysis in your inbox.
🚀 Stay ahead on AI
Get the latest AI news, tools and tips delivered to your inbox every week.
About the author: Flavien Hue has been testing and analyzing artificial intelligence tools since 2023. His mission: democratizing AI by offering practical and honest guides, without unnecessary technical jargon.