ChatGPT Visual Ads: OpenAI to Show Ads During Image Generation

ChatGPT visual ads are coming: OpenAI said on October 5, 2026 that it will start testing a new image-based ad format in the United States later this month, shown to users while ChatGPT is generating an image for them. The ads will be labeled, kept separate from the picture being made, and, according to OpenAI, will not influence ChatGPT’s answers.

I have been half-expecting this one since the first text ads showed up, and I still felt a small jolt reading the announcement. The few seconds you spend watching an image render are the last quiet corner of the product. It is the pause where you lean back and wait. OpenAI has looked at that pause and seen what every media company eventually sees in a pause: inventory.

What OpenAI actually announced

The details are fairly narrow, so it is worth being precise about them. According to BleepingComputer’s report on the announcement, OpenAI will begin testing a visual ad format this month with an initial group of advertisers, in the US only. The ads appear during image generation, which is the stretch of time between you asking for a picture and the finished picture arriving.

OpenAI describes the format as one that can “use images to show product inspiration, how a product is used, or experiences associated with a service.” So picture a furniture brand showing a styled living room, or a travel company showing a coastline, in a panel near the image you are waiting on. The company says the ads “will be clearly labeled, and remain separate from the image being created.” In other words, the sponsor does not get painted into your picture. Your request for a watercolor fox stays a watercolor fox.

Alongside the format itself, OpenAI gave a number that explains why advertisers care: ChatGPT now reaches 1.2 billion people every week. Advertising is framed as a second revenue stream next to paid subscriptions, not a replacement for them.

ChatGPT visual ads at a glance

Why the waiting moment is so valuable

Here is the analogy I keep coming back to. Think of the old cinema experience. Nobody bought a ticket for the ads, but everyone sat through them, because the lights were down, the seat was comfortable, and the film had not started yet. Attention was already gathered in one place with nothing else competing for it. Image generation creates the same condition in miniature. You have asked for something, you are curious about the result, and your eyes are fixed on one rectangle of screen.

That is a different kind of ad slot from a link at the bottom of a text answer. A text ad interrupts reading. A visual ad during rendering fills a gap that was otherwise empty, which makes it feel less intrusive even though it is arguably more attention-grabbing. I think that is exactly why OpenAI chose this spot for its first image-led format. It is the placement least likely to make people feel their answer has been tampered with.

It also fits what people use image generation for. Someone asking for a mood board for a kitchen renovation, a birthday invitation, or a logo concept is often close to spending money. Advertisers have always paid more to reach people at that point.

The plumbing matters more than the picture

The part of the announcement that got fewer headlines is the list of partners, and to me it says more about OpenAI’s ambitions than the ad format does. Hightouch, Tealium and LiveRamp are being connected for conversion measurement, the unglamorous work of letting a brand check whether someone who saw an ad later bought something. AppsFlyer, Adjust, Branch, Triple Whale and Kochava are in as attribution partners. DoubleVerify and Integral Ad Science have been added to pilot brand suitability evaluations, which is the industry’s way of checking that an ad did not appear next to something a brand would be embarrassed by.

If you have never worked near advertising, those names will mean little. If you have, you will recognize them as the standard toolkit a large advertiser expects before it moves serious budget onto a platform. A company running a small experiment does not wire up eight measurement and verification vendors. A company building a durable ad business does.

OpenAI has attached a privacy promise to this. It says the independent partners “will not get access to private user conversations” as part of those evaluations, and that safeguards are meant to keep ads out of sensitive conversations. I take that at face value for now, with the note that it is a claim from the company about its own system, and outsiders cannot yet check how well it holds.

How this fits the wider ChatGPT ads story

This is an expansion, not a debut. Reporting from The Next Web places the new format in a longer run: it says ads have been appearing in 31 European countries since late August, that free and Go tier users see them while higher paid tiers such as Pro and Enterprise do not, and that ChatGPT’s ad business passed 1 billion dollars in annualized revenue within 200 days. Those figures come from that outlet’s reporting, so treat them as reported rather than as numbers I can confirm independently.

The European angle is the one I would watch. We covered the moment the EU designated ChatGPT a very large online search engine under the Digital Services Act, and that label carries an obligation that suddenly looks very relevant: a public repository of the ads a platform has run, including who paid for them and how they were targeted. A US-only test of visual ads sidesteps that for the moment. A wider rollout would not.

There is also an uncomfortable contrast in the timing. The same company is, right now, holding back a model over safety concerns and dealing with scrutiny we wrote about when GPT-6 Astra crossed a critical cybersecurity threshold. Caution on the model side and acceleration on the revenue side are happening in the same month. That is not hypocrisy, since both can be true of a business that has enormous compute bills to pay, but it does tell you where the pressure is coming from.

What I would want answered before trusting it

My worry is not that ads exist. Free products used by more than a billion people are paid for somehow, and ads are a more honest arrangement than most alternatives. My worry is about the boundary OpenAI has promised to keep, and three questions sit on that boundary.

• What picks the ad? If the visual ad is chosen based on the image prompt you just typed, then your creative request is being read as a commercial signal. OpenAI has not spelled out the matching logic for this format in the coverage I have read.
• How is “sensitive” defined? Keeping ads out of sensitive conversations sounds right. The value of the promise depends entirely on where the line is drawn and who audits it.
• Does separation survive success? Today the ad sits apart from the generated image. If the format performs well, the commercial pull will be toward closer integration. The commitment that matters is the one made when the money is on the table.

For the record, I do not think the sky is falling. Search engines have run ads beside answers for a quarter of a century and remained useful. But a chatbot is a more intimate tool than a search box. People tell it things. The standard it has to meet is higher, and “trust us, the answers are not affected” will need to be backed by something people outside the company can inspect.

What you can do right now

If you use ChatGPT on a free plan in the US, you may start seeing a labeled visual panel during image generation at some point this month, though a test with an initial group of advertisers will not reach everyone at once. Nothing changes in how you write prompts. If you would rather not see ads at all, the reported exemption for higher paid tiers is the only route OpenAI has described, and I would check the current plan page before paying for that reason alone, because tier rules have shifted before.

You can read the original report on BleepingComputer for the full partner list and OpenAI’s wording.

Frequently asked questions

When do ChatGPT visual ads start?

OpenAI says testing begins later in October 2026, in the United States, with an initial group of advertisers. No date has been given for other countries or for a full rollout.

Will ads appear inside the images ChatGPT generates?

No, according to OpenAI. The company says ads will be clearly labeled and will remain separate from the image being created. The ad is shown during generation, not placed into your picture.

Do ads change what ChatGPT tells me?

OpenAI states that advertising does not influence ChatGPT’s answers. That is the company’s own assurance, and there is not yet an independent audit of it that the public can read.

Can advertisers see my ChatGPT conversations?

OpenAI says its independent measurement and brand suitability partners will not get access to private user conversations, and that safeguards are designed to keep ads away from sensitive chats.