ChatGPT ads are moving into images, so B2B teams need an AI visibility plan

What happened

On 5 October 2026, Engadget reported that OpenAI is testing more visual advertising in ChatGPT. The new ads will first appear when users generate images with ChatGPT Images. According to OpenAI's blog post, as quoted by Engadget, ads will be clearly labelled and remain separate from the image being created. OpenAI also states that advertising does not affect ChatGPT's answers.

The test begins later in October. US users will see the ads first, and only a select group of initial advertisers is taking part. ChatGPT Plus and Pro subscribers and Enterprise customers will not see ads, which places the format in front of users on the Free and Go plans.

The report adds some context. OpenAI began showing ads in ChatGPT at the start of 2026, in a sponsored section below responses showing the sponsor's name, logo and website link. Ads will not appear alongside sensitive topics, such as personal health or mental wellbeing. OpenAI plans to test additional ad formats.

Engadget's article does not cover measurement or pricing, and we have not seen OpenAI's own post, so we have limited this piece to what the report states.

Why it matters

The format is minor, and the direction is not. Ads next to generated images are unlikely to matter to a B2B services firm this quarter. What matters is that an AI assistant has become an advertising channel, with new formats on the way. Marketing teams that treat these assistants only as a content topic will soon be asked about them as a media line too.

Paid and organic presence are different jobs. OpenAI says advertising does not influence ChatGPT's answers. If that holds, buying a placement does not make an assistant recommend you in its responses. Being described accurately and favourably in answers depends on the material the assistant can find and trust about you: clear positioning, consistent facts across your own site and third-party sources, and content that answers real buyer questions.

Your measurement needs to cover both. Many B2B teams can report on search rankings and paid clicks but cannot say whether an assistant mentions them, or what it says. As ads and answers sit closer together on the same screen, that gap becomes harder to defend.

Audience fit decides the value. The first formats reach Free and Go users in the US. Enterprise buyers on business plans will not see these ads. For a UK, European or Gulf B2B audience, this is a signal to watch, not a budget line to add.

In our view, the sensible response is not to chase each new format but to build a base that works whatever the format turns out to be.

What we would do

1. Baseline your presence in AI answers. Write twenty to thirty questions your buyers actually ask. Run them in ChatGPT and the other assistants your audience uses. Record whether you appear, how you are described and which sources are cited.
2. Fix the facts first. If an assistant gets your services, locations or credentials wrong, correct the underlying sources: your own pages, your profiles and the third-party listings it relies on.
3. Publish answers, not slogans. Create pages that answer specific buyer questions plainly, with named authors, evidence and dates. These serve search, assistants and human readers alike.
4. Keep paid tests small and separate. If you test advertising inside an assistant, set a modest budget, a clear question to answer and a stop date. Track it separately from organic visibility so the two are not confused.
5. Check your tracking and consent set-up. Know what data you send to any advertising platform, on what legal basis, and who approved it. Take advice where personal data is involved.
6. Review monthly. Repeat the same question set each month. Changes in how you are described are an early indicator that is cheap to collect.

The teams that do well here will be those that treat AI visibility as a measured programme with an owner, not a reaction to each announcement.