ChatGPT Ads Go Live Across 31 European Markets Today — With a Two-Tier Privacy Rulebook Unlike Anything Before It

Today is the day Europe meets ChatGPT ads. On Monday, August 24, 2026, OpenAI begins serving sponsored messages to users across 31 European markets — including Germany, France, Spain, Italy, the Netherlands, Austria, Poland, Sweden, Norway, Denmark, and Ireland — covering the full European Economic Area in a single wave. It is the largest expansion of the company’s young advertising business to date, and it lands exactly at the six-month mark of an ads operation that started as a quiet U.S. pilot in February and has since marched through Canada, Australia, New Zealand, the U.K., Mexico, Brazil, Japan, and South Korea.

The scale is easy to state, but the interesting part is the legal architecture underneath. Europe is not getting a copy of the American product. It is getting something no major advertising platform has shipped before: a two-tier privacy rulebook in which the invasive layer is gated behind explicit consent, and the baseline layer runs without asking permission first.

What European users actually see today

Ads appear exclusively for users on the Free tier and the low-cost Go plan (€8/month). They surface below a response, labeled “sponsored” and visually separated from the model’s answer. Plus (€23/month), Pro, Business, Enterprise, and Education accounts remain ad-free.

At launch, every ad European users see is contextual only. Placement draws on five signals: the current conversation topic, approximate city-level location, device type, time of day, and language. Past chats and stored memories are explicitly excluded. For this processing, OpenAI relies on GDPR’s legitimate interest basis under Article 6(1)(f) — no consent banner, no permission request. The logic is defensible: session-level contextual targeting that doesn’t profile users across time is exactly the kind of narrow processing regulators have historically tolerated under legitimate interest.

The personalization layer is another story. For users who opt in through Settings → Ad controls, OpenAI can draw on a much richer data set: ad interaction history, inferred interests, and conversion data from advertising partners — what a user actually purchased after clicking. That layer requires an explicit, affirmative opt-in. Consent can be withdrawn at any time, and ad data can be deleted from the same settings panel, though OpenAI says removal from its servers can take up to 30 days.

Critically, declining personalization does not remove ads. It only changes which ads you see. To escape sponsored content entirely, a user must either accept a stripped-down ad-free Free variant (fewer daily messages, no image generation, no deep research) or pay for Plus. That structural reality — pay or see ads — is where the regulatory fight is likely to land.

Why Meta’s decade of GDPR pain forced a different design

To understand why OpenAI built it this way, look at what happened to its predecessor. Meta spent years arguing it could run behavioral advertising without asking permission — first claiming ads were part of its contract with users, then claiming legitimate interest. The Irish Data Protection Commission and the European Data Protection Board rejected both arguments and fined Meta €390 million in January 2023. Consent became Meta’s only lawful basis once every other door was closed.

OpenAI inverted the playbook. It starts from consent for personalization — the thing GDPR clearly requires — and reserves legitimate interest for the narrow contextual baseline that regulators have been most willing to accept. Whether that framing survives contact with the Irish Data Protection Commission, the lead supervisory authority over OpenAI Ireland Limited (the EEA data controller, based in Dublin), is an open question. The DPC has not yet publicly commented on the rollout or its privacy policy.

The timing is also sharper than it looks. The EU AI Act’s enforcement powers and fining authority became active on August 2, 2026 — 22 days before these ads reach European users. Italy’s Garante already fined OpenAI €15 million in December 2024 over ChatGPT’s data processing (the Court of Rome annulled it on jurisdictional grounds in March 2026, but the enforcement playbook is proven). Every European OpenAI launch now happens under active supervision with real financial stakes.

The business behind the rollout

The commercial engine behind this expansion is growing fast. OpenAI has reported 25% daily ad revenue growth since the start of August, and the ads business is approaching a $1 billion annualized run rate. The platform now supports CPC and CPM bidding, conversion optimization, geo-targeting, custom audiences, and third-party measurement via the OpenAI Pixel and Conversions API.

At Cannes this summer, OpenAI’s VP of Ads Dave Dugan — who only joined the company in March — said ChatGPT had passed 900 million weekly active users, with roughly 20% of queries showing direct commercial intent. “People come to ChatGPT with goals,” Dugan said in the expansion announcement. “In the attention economy, brands have to be interesting; in the intelligence economy, they have to be useful.”

European advertisers won’t get self-serve access at launch. Buying runs through the OpenAI Ads Solutions team and agency partners — Publicis, Omnicom, WPP, Havas, Dentsu, and MediaPlus — with self-serve Ads Manager access promised later this summer. (The U.S. self-serve tool launched this spring with a $50,000 minimum spend, since dropped.)

Agency practitioners describe a platform evolving at unusual speed. “The ad platform that we are using now is not the platform that was released as a pilot in April at all — it’s actually grown massively,” said Victor Batista, senior director of paid media at Jellyfish, whose clients were part of the pilot.

Guardrails: what ads won’t touch

OpenAI has drawn explicit exclusions. Ads won’t appear near sensitive or regulated topics such as personal health, mental health, or politics. Political advertising is banned outright. Temporary chats and the ChatGPT Atlas browser are excluded. And predicted minor accounts are removed from the ad system entirely — determined partly through age prediction that factors in stated age, account age, times of use, and discussed topics — a protection not publicly announced for the U.S. rollout.

On the model side, OpenAI insists advertising does not influence ChatGPT’s answers: ads run on separate infrastructure from the chat model, advertisers get only aggregate metrics (views, clicks), and no advertiser can pay to alter what ChatGPT tells a user.

Two structural questions remain open. First, the “consent or pay” problem: the EDPB’s Opinion 08/2024 held that platforms must offer a genuine, equivalent alternative without behavioral advertising — whether a feature-stripped free tier satisfies that is untested. Privacy attorney Clara Westbrook of Arbor Law put the pressure plainly: “It almost feels like people are having to pay if they don’t want to receive ads at all.” Second, the publisher question: ChatGPT’s answers are built from publisher content, and almost none of the new advertising margin flows back to the sources. Europe happens to be the world’s most aggressive jurisdiction for AI copyright enforcement.

The six-month-old ads business has now reached the market most likely to test its limits. Today the ads go live; the arguments about what they’re allowed to do follow close behind.