Markdown edition

# OpenAI plans visual ads during ChatGPT image generation

> OpenAI plans to test labeled visual ads during ChatGPT image generation in the US later in October. The company says ads will stay separate from generated images; the screen layout and user eligibility are unspecified.

By BIG CHANGE Editorial

Published: 2026-10-05T16:14:25.733Z
Updated: 2026-10-05T16:14:25.733Z
Canonical: https://bigchange.ai/blog/openai-chatgpt-image-generation-visual-ad-test

![Conceptual illustration of a large flower image canvas and a smaller product card on separate stands, with a clear gap between them.](https://bigchange.ai/api/media/file/openai-image-ad-boundary-hero-v1.png)
AI-generated conceptual illustration by BIG CHANGE.

OpenAI plans to test a visual ad format while people generate images in ChatGPT. In its [October 5 announcement](https://openai.com/index/new-chatgpt-ads-format-and-measurement/), the company said the ads would be labeled and kept separate from the image being created. Testing is due to begin later this month in the United States with an initial group of advertisers.

The distinction matters to anyone using ChatGPT to make an image: the advertisement is a paid placement in the creation experience, while the generated image is the work the user requested. OpenAI has described that boundary, but has not shown in the announcement exactly where the ad will appear on screen or which ChatGPT users will see this particular test.

## The big change

- **What changed:** OpenAI has announced a way to put visual advertising into an image-creation session. Its stated boundary keeps the paid placement separate from the generated image; the format is still awaiting a US test.
- **Why it matters:** For people making images, the location and label will determine whether a commercial message is clearly recognizable while they work. Advertisers gain a proposed placement tied to visual creation, but the announcement supplies no results for this format.
- **What to watch:** OpenAI has yet to specify the placement's exact presentation or eligible users. Those are the next details to verify as testing proceeds. Its new measurement integrations are available in its ad system now, while its work on experiments intended to measure incremental impact remains early.

## A visual placement with few interface details

OpenAI describes images that can show product inspiration, product use or experiences associated with a service. Its initial test will run during image generation. The company says the ad will be clearly labeled and separate from the image being created. It also says, as a statement about ChatGPT ads generally, that advertising does not influence the answers ChatGPT provides.

Those statements establish the proposed separation, not a demonstrated layout or independently measured behavior. OpenAI says the ad will appear during image generation, but the announcement does not show its exact location on screen or say whether a user can dismiss it. It also does not specify eligibility by subscription tier for this new visual format. OpenAI's [January advertising principles](https://openai.com/index/our-approach-to-advertising-and-expanding-access/) described an earlier US test aimed at logged-in adults on the Free and Go tiers, with ads separated from answers. That earlier scope should not be assumed for the visual test without a further statement from OpenAI.

## Measurement grows before the visual test starts

The same October announcement expands the advertiser side of ChatGPT Ads. OpenAI says integrations with Hightouch, Tealium and LiveRamp make it easier to send conversion data from advertisers' existing systems into ChatGPT Ads. It also lists partners for attribution, reporting and experiments intended to assess whether ads cause additional business outcomes. OpenAI says its work on that last question, known as incrementality, remains at an early stage.

This measurement expansion and the planned image-generation test are different developments. OpenAI cites early results from partners for ChatGPT Ads, but those examples are not presented as results from the visual format announced on October 5. They cannot show how people will respond to a placement that has yet to begin testing.

For users, the next evidence to inspect is the labeled placement when testing proceeds. For advertisers, the practical question is what the new format can measure once campaigns run. OpenAI has announced the boundary between ad and image; its on-screen presentation remains to be seen.

## Sources & further reading

- [OpenAI, “Building advertising for the way people use AI” (October 5, 2026)](https://openai.com/index/new-chatgpt-ads-format-and-measurement/). The primary announcement for the planned US visual-format test, its stated labeling and separation, the initial advertiser group and the measurement partnerships. It describes company plans and claims; it does not document a live user test of this format.
- [OpenAI, “Our approach to advertising and expanding access to ChatGPT” (January 16, 2026)](https://openai.com/index/our-approach-to-advertising-and-expanding-access/). Sets out the company's answer-independence and ad-separation principles and the scope of its earlier ad-test plan. It does not specify eligibility for the new image-generation format.

## Sources

- [Building advertising for the way people use AI](https://openai.com/index/new-chatgpt-ads-format-and-measurement/) — OpenAI's primary announcement says it plans a labeled visual ad test during ChatGPT image generation later in October in the US with an initial group of advertisers. It says ads will remain separate from generated images and describes measurement integrations. It does not establish that the visual format is live or show its precise layout or user eligibility.
- [Our approach to advertising and expanding access to ChatGPT](https://openai.com/index/our-approach-to-advertising-and-expanding-access/) — OpenAI's earlier ad principles include answer independence, labels and separation, plus the stated scope of an earlier US Free and Go test. They do not establish eligibility for the newly announced visual format.