ChatGPT's ads are coming to Europe — and the GDPR fine print, not the ads themselves, is what should change how you use it

What OpenAI actually told European users

At 08:29 GMT on 15 August 2026, OpenAI Ireland Limited — the entity that controls the service for European users — sent a notice to Free and Go subscribers across the European Economic Area and Switzerland telling them that ads will begin appearing in ChatGPT “later this month.” That single email is the news. It is not a new product; it is the moment an ad test OpenAI has been running since a US pilot on 9 February 2026 finally reaches the one jurisdiction where the rules governing it are strictest.

The rollout sequence, confirmed across OpenAI’s own materials and reporting from MLQ News and Tuta, runs like this: the US pilot opened in February, self-serve US advertisers were let in during May, and on 11 August OpenAI expanded the test to five more countries — the UK, Mexico, Brazil, Japan and South Korea. (Reporting varies on the exact prior dates for some of those markets; the direction — US first, then a widening ring — is not in dispute.) Europe’s regulated markets are the meaningful new frontier, because in the EEA and Switzerland the question is no longer “will people tolerate ads” but “is this lawful under the GDPR.”

The mechanics are the reassuring part

Taken at face value, OpenAI’s design is conservative, and worth stating precisely because the details are where buyers make decisions:

• Who sees ads. Only Free and Go users. Plus, Pro, Business, Enterprise and Education accounts stay ad-free. That boundary matters more than it looks — it converts “no advertising” into an explicit, priced feature of every paid plan.
• Where they appear. Below ChatGPT’s responses, not woven into them.
• How they’re targeted. Initial selection uses the current conversation topic plus limited context such as general location and device type. Past chats and saved memories are excluded from that initial selection. Personalized advertising — which draws on the ads you’ve interacted with and context pulled from your chats — requires a separate opt-in.
• What OpenAI promises. Ads “do not influence the answers ChatGPT gives you,” and conversations are “kept private from advertisers.” Users retain an “unconditional right to object” to processing for direct marketing.

If those commitments hold, the day-to-day experience for a casual European user is a contextual ad under some answers and nothing else. There is no evidence at launch that answers are being steered toward advertisers, and OpenAI has drawn the sponsored-content line more carefully than the search engines did a generation ago. The honest caveat: these are OpenAI’s stated policies, not independently audited facts. “Ads don’t influence answers” is a claim that is very hard for an outside party to verify on a closed model, and it is exactly the claim worth watching as the ad business grows and the incentive to blur the line grows with it.

The part that isn’t settled: consent-or-pay under the GDPR

Here is where a routine monetization story becomes a European legal one. To let users avoid ads without upgrading to a paid tier, OpenAI is offering an “ads-free free plan” — free of ads, but with a reduced message limit. On its face that sounds generous. Under EU data-protection law it is one of the most contested designs in the field.

Offering “accept ads, or accept a degraded free service” is a consent-or-pay model — the same “pay-or-okay” pattern European publishers have deployed and regulators have spent two years pushing back on. The European Data Protection Board’s Opinion 08/2024 concluded that, in most cases, large online platforms cannot obtain valid, freely given consent by putting users to a binary choice between paying (or here, accepting a worse product) and consenting to behavioural advertising. The reasoning: if the alternative to consent is meaningfully worse, the consent isn’t truly free. This is not theoretical — it is being enforced. A live Norwegian action against Schibsted’s pay-or-okay system (a complaint filed on 3 June 2026) is the kind of case that will shape whether OpenAI’s European structure survives contact with a regulator.

That puts OpenAI’s “ads-free free plan with fewer messages” in an awkward spot. If a data-protection authority decides the reduced-message tier is a degraded alternative rather than a genuine equivalent, the consent underpinning the ad targeting could be challenged — and OpenAI would be pushed toward either a truly unconditional free ad-free tier or a paid-only escape from ads. For readers weighing the EU AI Act’s transparency regime, this is the same theme from the privacy side: Europe is where the AI industry’s default business models get stress-tested first.

Where your data goes

The second European wrinkle is transfer. OpenAI’s notice states that 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) and Standard Contractual Clauses under Article 46(2)(c) — the standard legal machinery for moving personal data across the Atlantic. There’s nothing exotic here; most US cloud services rely on the same instruments. But it does mean the plain-language description of the free tier is now: your conversation topics are read by an ad-selection system, and the data underpinning it leaves the continent. That is a fact worth internalizing before typing anything you wouldn’t want in an ad pipeline.

Why this matters

The free tier’s price was never zero — now it’s explicitly ad-funded. This announcement doesn’t stand alone. In the same fortnight, OpenAI made GPT-5.6 Luna the default for Free and Go users with unlimited text chats after cutting Luna’s API price ~80% amid a broader price war, and Google’s Gemini crossed one billion monthly users. Read together, the message is unambiguous: the major labs are racing to own the free consumer tier, and OpenAI has now told you how it intends to pay for an unlimited free product used by hundreds of millions — with ads. Free got bigger and cheaper to use precisely because it’s becoming an advertising surface. That’s not a criticism; it’s the business model finally saying its name.

“No ads” is now a concrete, priced feature — factor it into the buy. Because every paid tier stays ad-free, the decision to pay for ChatGPT Plus (or to standardize on a paid competitor — the Gemini-vs-ChatGPT call now carries an ads asterisk) now includes “keep an ad-selection system out of my conversations” as a line item, not a vibe. For individuals that’s a $20/month question; for teams it’s a procurement one. If your organization has people doing real work in free ChatGPT, the ad rollout is the prompt to move them onto a paid or enterprise tier — or off it.

For sensitive, client or regulated work, the free tier is now clearly the wrong tool. It arguably always was, but ads make the data flow explicit: conversation topics feed targeting, and processing crosses borders under SCCs. Keep confidential material out of the free tier — and, honestly, out of any consumer chatbot tier — and lean on Claude, enterprise ChatGPT, or another ad-free, contractually-bounded deployment for anything you’d be uncomfortable seeing in an ad system. The chatbot landscape now has an “ads on free tier?” axis, and it’s a legitimate one — Perplexity already runs sponsored follow-ups, so the honest comparison is not “ads vs no ads” but “which free assistant’s ad model you trust with your topics.”

Watch the consent-or-pay challenge, not the ad units. The ads themselves are the least interesting part of this story. The interesting part is whether Europe forces OpenAI to offer a genuinely free, genuinely ad-free tier — the outcome the EDPB’s reasoning points toward. If a regulator lands that ruling, it resets the economics of free AI assistants across the continent, and every rival watching this rollout will have to answer the same question.

The verdict

For a casual European user, the practical impact of this change is small: contextual ads under some answers, off the paid tiers entirely, non-personalized unless you opt in, and — per OpenAI — with no influence on the answers themselves. Tolerable, and more restrained than the ad-tech norm.

The recommendation for anyone using ChatGPT for anything that matters is sharper. Treat the free tier as an ad-supported consumer product, because that is now what it is: fine for casual questions, wrong for work, client, or sensitive data. If ad-free handling matters to you, that guarantee lives on the paid tiers — or with a competitor that doesn’t monetize its free plan this way. And keep an eye on Brussels and Oslo rather than the ad slots: the durability of OpenAI’s European model rests on a consent-or-pay question the EU has already signalled it intends to answer, and that answer will shape how every free AI assistant is funded next.

Questions or corrections? Email Pick Right. Want the full list? See all news.