ChatGPT Ads is here! 5 things I think OpenAI got right

OpenAI has started testing ads in ChatGPT. From today, logged-in adult users on the Free tier and the Go plan in the US will see ads below answers. Plus, Pro, Business, Enterprise and Edu stay ad-free.

It was expected, free users cost enormous amounts of compute. What is interesting is not that ads are coming, but how. And here I think OpenAI has done several things better than one might have feared.

Five things I think OpenAI got right

1. Ads do not influence answers

OpenAI calls it answer independence: ads run on separate systems from the chat model, and advertisers cannot buy their way into the answer itself. This is the most important line of all. An answer you cannot trust is worthless, to the user and, eventually, to the advertiser.

2. The ad is clearly an ad

Ads sit below the answer, are clearly labelled and kept visually separate. No sponsored sentences slipped into the text. It sounds obvious, but it is exactly the line search ads have blurred over time.

3. Conversations are not sold

Advertisers do not get access to conversations or personal details. They get aggregate performance data, not what you wrote. That is a stronger promise than much of the ad-funded web can make.

4. Users stay in control

You can turn off personalisation, see why an ad is shown and dismiss ads you do not want. Anyone who wants no ads at all can pay for Plus. The choice is clear and not hidden in settings.

5. Sensitive topics and minors are kept out

No ads are shown to accounts judged to belong to people under 18, and ads are kept away from sensitive topics such as health and politics. It is a clear way of saying some conversations are not ad space.

How ChatGPT ads work

The ads system uses the context of the current conversation to pick an ad that fits what you are exploring. Ask about a renovation and you might see a hardware retailer; compare business software and a vendor may appear below the answer. Depending on country and settings, the system may also use your broader ChatGPT history, if you allow it.

The difference from Google is where in the journey the ad lands. In search, the ad appears after a couple of words. In ChatGPT you have often already described your goal, requirements and constraints. The ad lands in the middle of the decision, not at its start.

In the test, OpenAI works with a small number of advertisers. Delivery is fully controlled by OpenAI's system, the advertiser chooses message and budget, but not which answer the ad appears under.

What it means for marketers

Three things. First: the ad is no shortcut into the answer. The organic answer is still decided by what the models have learnt about your category, so visibility in AI answers matters just as much. Second: relevance beats reach. An ad below a well-formed decision brief must be a useful next step, not an interruption. Third: start preparing messages per buying situation now, even before ads reach your market.

Update, October 2026

A lot has happened since the test began. In May, OpenAI opened a beta self-serve Ads Manager and introduced cost-per-click (CPC) bidding, with agencies such as Dentsu, Omnicom, Publicis and WPP as partners. Through spring and summer ads rolled out in markets including Canada, Australia, the UK, Japan and Brazil. On 31 August OpenAI said ChatGPT Ads had reached a $1 billion annualised revenue run rate and that advertisers can now buy ads via Ads Manager in Europe too. The figures come from OpenAI itself.