OpenAI today told European ChatGPT users that advertising will begin appearing on the Free and Go plans later this month, and that its privacy policy is being updated to describe how those ads are selected, measured and controlled.

The notification arrived by email at 08:29 on Saturday, August 15, 2026, sent from noreply@email.openai.com under the subject line "Updates to OpenAI's Privacy Policy" and signed by the OpenAI Team rather than by any named executive. Its footer carries the address at 1st Floor, The Liffey Trust Centre, 117-126 Sheriff Street Upper, Dublin 1, D01 YC43 - the registered office of OpenAI Ireland Limited, which the company's European privacy policy names as the controller responsible for processing the personal data of users in the European Economic Area and Switzerland.

That address matters more than a standard email footer normally would. Six months after ChatGPT advertising launched in the United States, the continent that OpenAI had conspicuously skipped is now inside the map.

What the notice says

The commercial terms are set out in a single paragraph. Ads may appear on Free and Go plans beginning later this month, according to OpenAI. Plus, Pro, Enterprise, Business and Education plans do not have ads. The company describes ad-free options as remaining available.

Selection, at the outset, is contextual rather than behavioural. "To begin, ads will not be personalised," the email states. The signals named are the current topic of the ChatGPT conversation and limited contextual information such as general location and device type. Past chats and memories are excluded from that initial selection process.

Personalisation is framed as a later step gated by an explicit prompt. According to the notice, before personalised advertising begins the company will ask each user to choose whether to opt in, and participation would draw on information described as staying only on ChatGPT, including ads the user has interacted with and context drawn from chats. Ad personalisation is to be manageable in settings.

Measurement is described in narrower terms than the industry standard. Information about ads viewed or interacted with is used to measure and improve effectiveness, and the email states that "Advertisers only receive overall information about how their ads perform, such as total views or clicks."

Under the heading of what is not changing, OpenAI repeats two of the five advertising principles it published in January: "Ads do not influence the answers ChatGPT gives you," and that ads are clearly labeled and visually separated from responses. Advertisers, the notice adds, have no access to chats, chat history, memories or personal details.

The selection machinery underneath

OpenAI's help centre article on ads in ChatGPT, updated within hours of the European email going out, sets out the mechanics in more detail than the notification does.

The placement is fixed: ads can appear below the end of a response, labelled as sponsored and visually separated from the answer. During the test, ads do not appear in the ChatGPT Atlas browser, and Temporary Chats do not show ads at all.

The selection logic considers several inputs at once. According to OpenAI, the ads system weighs the context and intent of the current conversation, the ad's landing page, its title and copy, advertiser-supplied context hints and targeting selections, and - where personalisation is enabled - selected signals from the wider ChatGPT experience. Ranking across multiple eligible ads combines relevance with advertiser bids. That is an auction, described in plain language.

One line in the help centre resolves the apparent gap between a global product and a staged European launch. "Personalized ads are not initially available in the European Economic Area (EEA) or Switzerland," the document states, adding that ad controls themselves are available only to Free and Go users in regions where ads are available or will soon be available. The email's promise of an opt-in prompt and the help centre's regional carve-out describe the same sequencing from two directions.

Controls sit in Settings under Ad Controls. The panel exposes viewing history and the topics used to make ads more relevant, a delete function for ads data, a personalisation toggle, and a separate toggle governing whether past chats and memory contribute to relevance. Deleted ads data is retained for up to 30 days before removal from OpenAI's servers. Turning personalisation off does not remove advertising; it narrows selection back to the current chat thread.

Individual units carry a three-dot menu offering Hide ad, Report ad, About this ad, and an Ask ChatGPT function that shares the specific creative into the conversation. By default, according to the documentation, the model cannot see the ads served alongside its own output.

Where ads are excluded

Two exclusions are categorical. Accounts identified as belonging to users under 18 do not receive ads, and the identification method is disclosed in unusual detail: account-level age information plus age prediction that weighs stated age, how long the account has existed, typical times of use, usage patterns over time, and the general topics a person discusses. Behavioural inference is being used here to withhold advertising rather than to target it.

The second exclusion covers subject matter. "Ads are not eligible to appear near sensitive or regulated topics, including personal health, mental health, or politics," according to the help centre. Advertisers in regulated verticals such as health and financial services may qualify under stated eligibility criteria, but placement near sensitive user content is excluded regardless, and political advertising is not currently permitted anywhere in ChatGPT.

What the privacy policy currently contains

Three versions of OpenAI's European privacy policy were reviewed alongside the notification: an archived text dated November 4, 2024, a version dated February 6, 2026, and the version dated June 4, 2026.

None of them mentions advertising.

The June 2026 European policy lists seven processing purposes in its legal bases table: providing, analysing and maintaining the Services; personalising and customising the experience; improving and developing the Services and conducting research; communicating with users and promoting OpenAI's own Services; identifying contacts who use the Services; preventing fraud and misuse; and complying with legal obligations. The personalisation purpose, which did not exist in the November 2024 text, is illustrated with Memory, Custom Instructions, topic suggestions and Sora video recommendations. Its stated legal bases are contract performance and legitimate interests. Advertising appears nowhere in that column.

The gap is the point. OpenAI's US privacy policy was rewritten on April 30, 2026 to state in binding language that the company receives purchase data from advertisers and their partners, shares user information with marketing partners for third-party targeting, and uses personal data to promote its own products. The European text has carried no equivalent disclosure. Today's email announces that this is changing, though the amended European policy itself was not published alongside the notification, and its effective date remains unstated.

Other provisions in the June 2026 text bear directly on how an ad business would operate under it. Deletion requests are honoured within 30 days. Temporary Chats are automatically deleted within 30 days. Processing takes place on servers outside the EEA, Switzerland and the UK, including in the United States, relying on European Commission adequacy decisions under Article 45(1) GDPR and on Standard Contractual Clauses under Article 46(2)(c). Users hold an unconditional right to object to processing for direct marketing, and a right to withdraw consent where consent is the basis relied upon.

That last pair of rights is why the sequencing described in the email is legally consequential rather than merely polite. Contextual selection from a live conversation, general location and device type is a materially different processing operation from profiling built on chat history and memory, and the two do not sit comfortably under the same lawful basis.

OpenAI enters this market with a structure that European regulators have spent two years arguing about in other contexts.

Users who prefer no advertising have two routes. One is a paid subscription: Plus, Pro, Business, Enterprise or Edu. The other is an Ads-Free configuration of the Free plan, which removes advertising in exchange for lower message limits and reduced feature access, with image generation and deep research cited as examples of tools that become unavailable. Switching runs through Settings, Ads controls, and a Change plan option, and the interface presents the trade-off explicitly before confirmation. Rate-limited users are then offered a banner to upgrade or to switch back to the ad-supported experience.

The presence of that third door distinguishes the arrangement from the models the European Data Protection Boardexamined in Opinion 08/2024, which found that most consent-or-pay implementations fail the GDPR standard for freely given consent and recommended that platforms offer a genuinely free alternative without behavioural advertising. Whether reduced message limits and withdrawn tooling amount to a genuinely equivalent free alternative, or to a degraded one that functions as pressure, is precisely the question that has kept publisher paywalls in front of regulators. A formal complaint against Schibsted's pay-or-okay system was filed with Norway's Datatilsynet on June 3, 2026, and research published in July 2025 found consent rates of 99 to 99.9 percent across European sites deploying payment barriers.

Jurisdiction over any complaint would run through Dublin. The Irish Data Protection Commission is OpenAI Ireland's lead supervisory authority, a position formally recognised in February 2024, and the consolidation effect of that recognition on parallel national proceedings has already been tested in litigation over the Italian Garante's fifteen million euro decision against the company.

Six months of groundwork

The European notice closes a gap that has been visible in OpenAI's own hiring since spring. When the company listed 19 advertising roles on May 6, 2026, the locations were San Francisco, New York, London, Tokyo, Seoul, Sydney and Sao Paulo. Paris, Amsterdam and Berlin did not appear. When the five-market expansion was announced on May 7, continental Europe was again absent.

The sequence elsewhere moved quickly. Advertising launched in the United States on February 9, 2026 for logged-in adults on Free and Go, at a 60 dollar CPM with a minimum commitment between 200,000 and 250,000 dollars. Criteobecame the first ad tech partner on March 2. The pilot crossed 100 million dollars in annualised revenue within six weeks, disclosed on March 26 alongside expansion into Canada, Australia and New Zealand. CPMs fell to as low as 25 dollars by mid-April. The self-serve Ads Manager opened to all US businesses on May 5, the minimum spend disappeared, and cost-per-click bidding arrived at a recommended 3 to 5 dollars. Custom audience targeting surfaced on May 14, daily budgets and geo-targeting on May 22, conversion-optimised campaigns on June 5, product feed support in June.

The United Kingdom went live on June 6, becoming the first European market to receive ChatGPT ads under a post-Brexit regime that operates outside EU GDPR enforcement. Japan and South Korea followed in July, with UK advertisers gaining self-serve access at the same time. More recently, fixed daily budget caps were converted into seven-day averages without an opt-out, and automatic advanced matching is scheduled to become the default on existing web pixels on August 17, 2026. AdRoll opened gated ChatGPT access to selected small and mid-market customers on August 11.

Against that pace, the European approach reads as deliberately slower: contextual first, personalisation held back, consent solicited separately.

Why this matters for the marketing community

For media buyers, the immediate consequence is inventory that behaves differently from every other European channel at launch. With personalisation withheld, matching depends on advertiser-supplied context hints against live conversation text, the ad's own landing page and copy, general location and device type. Creative and hint quality carry the entire targeting burden. Campaign structures tuned on US behavioural signals will not transfer cleanly.

For measurement teams, the reporting floor is aggregate views and clicks, with the help centre noting only that additional measurement insights may be explored over time. Automatic advanced matching, which uses hashed customer information collected from website forms, becomes the pixel default on August 17 in markets where it operates - a technique whose application under European law depends on consent captured at the advertiser's own form, not at OpenAI's.

For anyone modelling reach, the addressable base is narrower than ChatGPT's headline user count suggests. Paid tiers above Go are excluded by design. Under-18 accounts are excluded by prediction. Temporary Chats and the Atlas browser are excluded by product. Sensitive verticals are excluded by policy. And a free user who prefers fewer messages to more advertising can leave the pool entirely with five taps.

The documents released today do not confirm when, or whether, European advertisers will be able to buy this inventory directly. The help centre points to a separate availability page for advertiser access by country and notes that availability may continue to evolve as testing expands. The user-facing rollout and the buy-side rollout are being announced on different clocks, as they were in every prior market.

Timeline

Summary

Who: OpenAI, through OpenAI Ireland Limited, the data controller for users in the European Economic Area and Switzerland, notifying ChatGPT users on the Free and Go plans. Plus, Pro, Enterprise, Business and Education subscribers are unaffected.

What: Advertising is being introduced to ChatGPT in Europe, accompanied by a privacy policy update describing ad selection, measurement and controls. Selection begins without personalisation, using the current conversation topic plus general location and device type, and excluding past chats and memories. An opt-in prompt is to precede any personalised advertising. Advertisers receive aggregated performance data only, with no access to chats, chat history, memories or personal details.

When: The notification was sent at 08:29 on August 15, 2026, with ads described as appearing later in the same month. The advertising pilot has run in the United States since February 9, 2026, reached the United Kingdom on June 6, 2026, and Japan and South Korea in July 2026.

Where: The European Economic Area and Switzerland, governed by the European privacy policy administered from OpenAI's Dublin office, with the Irish Data Protection Commission as lead supervisory authority.

Why: OpenAI states that advertising supports broader access to ChatGPT and continued investment in the product while keeping the Free and Go plans fast and reliable, and that ads do not influence model answers. The staged approach in Europe, with contextual selection first and personalisation deferred behind an explicit choice, reflects a regulatory environment in which the lawful basis for behavioural advertising, the validity of paid opt-outs, and the adequacy of free alternatives are all actively contested.