How the New Format Works
OpenAI introduced a new visual ad format for ChatGPT in an official announcement on October 5, 2026, titled "Building advertising for the way people use AI." The new format appears on screen while the user is generating an image — the ad is tied to the generative image-creation process itself and placed in the image-generation window, not inside text responses.
The pilot begins in late October in the US: a limited group of advertisers will take part in the first stage. The format will first be tested within this narrow circle, after which expansion will be considered. The announcement's headline itself signals that the company is building advertising not around traditional banner logic, but around how people actually use AI.
The company highlighted two principles of the new format. First, ads will be clearly labeled — users will be able to tell an ad element apart immediately. Second, ads are shown separately from the image being created, so commercial content does not blend with the user's own generation.
According to OpenAI, ads will be clearly labeled, shown separately from the generated image, and will not affect ChatGPT's responses.
The third guarantee is that ads will not influence ChatGPT's responses. This statement means model answers will not depend on commercial interests: the answer a user receives will not change based on whether an advertiser paid or participated. For OpenAI this is a matter of trust — as advertising becomes a revenue source, questions about the neutrality of answers will naturally grow louder.
Why Now: 1.2 Billion Weekly Users
According to OpenAI, ChatGPT now reaches 1.2 billion people every week. That is one of the largest digital audiences in the world. At this scale, even a small share of ad impressions means a massive revenue stream — which is exactly why advertising is becoming one of the main ways to earn from users who don't pay for subscriptions.
Until now, the company earned most of its revenue from paid plans: ChatGPT Plus and Pro subscriptions and enterprise plans were the backbone of the business. Now the giant stream of free users is turning into a revenue source too. OpenAI says it openly: advertising will be a way to earn specifically from users who don't buy a subscription.
This direction did not appear out of nowhere. ChatGPT's pricing system has been changing actively in recent months: as written in our ChatGPT subscription plans review, adding ads to the free plan, the cheaper budget Go plan, and the two-tier structure of the Pro plan are all part of one process. The new visual format is the next step of this advertising infrastructure — ads will now appear not only as text, but in the image-generation interface as well.
Measurement Tools and Partners
Alongside the new format, OpenAI is expanding measurement tools for ChatGPT Ads. The most important question for advertisers — does the ad deliver real results — is why the company announced partnerships across three areas.
First, integrations with Hightouch, Tealium, and LiveRamp are being set up for conversion data. These platforms make it possible to connect ad impressions with a business's own data. Second, the list of attribution partners has grown: AppsFlyer, Adjust, Branch, Triple Whale, and Kochava will help determine which channel brought the customer. Third, work continues on brand safety with DoubleVerify and Integral Ad Science — advertisers' ads are monitored so they don't appear next to content brands don't want. Each partner on the list closes a separate link in the measurement chain: who saw it, who clicked, who bought — these questions will be answered.
On the same day, DoubleVerify announced important news for advertising in ChatGPT in a separate press release: the company launched attribution measurement for ChatGPT ads through DV Rockerbox.
According to DoubleVerify, the company is working with OpenAI on a pilot project to assess brand suitability without accessing users' private conversations.
This detail matters for privacy: brand suitability is assessed, but private conversations are not read to do it. As the advertising ecosystem expands, balancing user privacy with advertisers' measurement demands is becoming one of OpenAI's main tests.
Where the Pilot Launches
The pilot will first run in the US with a limited group of advertisers. On the very day the announcement came out, independent outlets covered it: BleepingComputer on how the format works and its guarantees, DoubleVerify on the details of the measurement partnership in its own press release.
You can read the official announcement on the OpenAI website; BleepingComputer also covered the details.
