You ask ChatGPT to help you pick a name for your project. The answer comes back, and right next to it sits a sponsored box for a web hosting service. That is the scenario waiting for French users of the free version: ChatGPT will start showing ads in France from late August 2026.
The change affects the Free and Go plans. OpenAI made these ads official in Europe through a new privacy policy published on August 14, 2026, covering the European Economic Area, the United Kingdom and Switzerland. The Plus, Pro, Business, Enterprise and Education subscriptions carry no ads.
Here is what actually appears inside a conversation, what the targeting is built on, what advertisers get to see of your messages, and the three ways to opt out. No advertising jargon required to follow along.
What an ad looks like inside ChatGPT
The ad shows up as a box placed beside the answer, labelled as sponsored. OpenAI says it keeps a clear visual break between that box and the text the model writes.
The choice of ad does not rest on your age or on a declared profile. It rests on what you are doing right now. OpenAI matches the ads submitted by advertisers against three things:
- the topic of the conversation in progress;
- your past chats;
- your earlier interactions with ads.
When several advertisers target the same topic, OpenAI picks the one it judges most relevant to the discussion. The example OpenAI gives in its announcement on testing ads: looking up recipes can trigger an ad for a meal kit or a grocery delivery service.
This flips the logic of Google Ads or Meta Ads. You no longer buy a keyword or a demographic audience, but an intention spoken in plain language inside a private conversation.
Who sees ads and who does not
The dividing line is not between free and paid. It runs between the entry-level plans and the rest of the catalogue.
| Plan | Ads | Personalisation |
|---|---|---|
| Free | Yes | Yes with consent, generic otherwise |
| Go | Yes | Yes with consent, generic otherwise |
| Plus and Pro | No | Not applicable |
| Business, Enterprise, Education | No | Not applicable |
Go is a paid subscription, and it shows ads. Paying is therefore not enough to escape them: check which tier you signed up for on the ChatGPT pricing page before assuming you are out of scope.
The two-tier European setup
The General Data Protection Regulation (GDPR, the European text that has governed the use of personal data since 2018) forbids personalising ads without a solid legal basis. OpenAI answers with a two-tier system.
First tier: personalised ads rest on your explicit consent. You accept, and your conversations feed the targeting.
Second tier: if you refuse, ads still appear, but in a generic form. OpenAI then relies on the legitimate interest basis described by France's data protection authority, which allows processing without prior agreement when a company can show a proportionate interest.
Refusing the targeting does not make the box disappear. It only changes how the box gets filled.
On the compliance side, OpenAI names OpenAI Ireland Limited as the data controller for the European Economic Area and Switzerland. European data protection authorities, including the CNIL in France, therefore have an identified counterpart. Using legitimate interest for non-personalised ads is the piece most likely to be challenged.
What advertisers see of your messages
OpenAI is explicit here: advertisers have no access to your conversations, your history, your memories or your personal information. They receive aggregated statistics, such as the number of views and the number of clicks.
The nuance matters. Your messages do not travel out to an advertiser, but they do decide which ad you see. The content of your conversation stays with OpenAI and drives the selection.
Two safeguards were announced when the test started. No ads on accounts declared as under 18, or believed to belong to a minor. And no ads near sensitive or regulated topics such as health, mental health or politics.
OpenAI writes that it wants to widen this scope "responsibly" as protections mature. The categories excluded now may not stay excluded: this is not a permanent guarantee, it is the current state of a programme still being built.
The three ways to stop seeing ads
OpenAI describes several controls in the account settings: dismissing a specific ad, sending feedback, checking why it was shown to you, deleting your advertising data, and managing personalisation. Consent can be withdrawn at any time.
To get rid of the ads themselves, there are three routes:
- move to a Plus or Pro plan, which carries none;
- turn ads off inside the free plan, accepting a smaller daily message allowance;
- do nothing beyond switching off personalisation, which leaves generic ads in place.
The second route is the most interesting and the least discussed. It turns advertising into an open trade: either you look at ads, or you talk less with the model.
How OpenAI got here
OpenAI kept advertising off the table for a long time. The programme was built in waves, market by market, across 2026.
- February 9, 2026: first tests in the United States, on adult Free and Go accounts.
- March 26, 2026: expansion announced for Canada, Australia and New Zealand.
- May 7, 2026: OpenAI says it is aiming for the United Kingdom, Mexico, Brazil, Japan and South Korea.
- August 11, 2026: actual launch in those five countries.
- August 14, 2026: new privacy policy for Europe, with ads announced for the end of the month.
To justify scaling up, OpenAI shared early results from its American test in March 2026: no impact on its user trust metrics, and a low rate of ad dismissals.
Both results were announced without numbers. We know neither the dismissal percentage, nor how the trust metrics are measured, nor the comparison baseline. A "low rate" with no absolute value cannot be checked independently.
For businesses, OpenAI collects sign-ups on its advertisers page, which also serves to get notified when new markets open.
What this actually changes for you, even working solo
Reread what you paste into ChatGPT before hitting Enter. The topic of your conversation decides which ad appears. If you work on a client file in the free version, that content now feeds a commercial mechanism. Sensitive client data has no business in a window funded by advertising.
Weigh three costs instead of letting the change happen to you. A Plus or Pro subscription costs money. Turning ads off in the free plan costs messages. Accepting the targeting costs attention and a slice of privacy. Make the call once, knowingly, rather than clicking through the first consent window you meet.
Start seeing ChatGPT as a place where your competitors can buy a seat. A plumber, an agency or a trainer can pay to appear next to an answer about their market. Entry budgets stay low in the early life of an ad network, before bidding heats up. If your customers find you through search, this is the moment to test with small sums.
Frequently asked questions
Who is affected by ads in ChatGPT in France?
Users of the Free and Go plans, from late August 2026. The Plus, Pro, Business, Enterprise and Education subscriptions show no ads. The scope announced by OpenAI on August 14, 2026 covers the European Economic Area, the United Kingdom and Switzerland.
Can you turn ads off in ChatGPT?
Yes, in three ways. By moving to a Plus or Pro plan, which serves none. By switching ads off inside the free plan, in exchange for a smaller daily message allowance. Or by refusing personalisation only, which leaves generic ads chosen without your data.
Can advertisers read my ChatGPT conversations?
No, according to OpenAI. Advertisers have no access to conversations, history, memories or personal information. They receive aggregated performance data on their ads, such as views and clicks. The content of the conversation is, however, used by OpenAI to decide which ad to show.
Do advertisers influence ChatGPT's answers?
OpenAI says they do not: answers are optimised only for how useful they are to the user, and the sponsored box is labelled and visually separated from the generated text. That separation rests on OpenAI's own word, and no external audit of the system has been published.
How can a freelancer run ads inside ChatGPT?
Sign-ups go through OpenAI's advertisers page, at openai.com/advertisers, which also serves to get notified when new markets open. OpenAI says it wants to gradually widen the formats, campaign objectives and buying models offered to businesses.
