OpenAI is building image, video, native, and conversational ad formats for ChatGPT — the first serious ad surface inside an AI assistant. Here is how it differs from social advertising, why the placement matters more than any single format, and the two things it asks of you: visual ad creative and a presence worth citing.
For twenty years, digital ads lived in two surfaces: search results and social feeds. In 2026 a third one is opening. OpenAI is building image, video, native, and conversational ad formats for ChatGPT, according to engineering job listings reported by Digiday and Search Engine Journal on July 1, 2026 — a step beyond the single text-and-image sponsored unit it has been testing. There is no launch date; the hiring signals intent and staffing, not a shipped product. But the direction is unambiguous, and it is worth understanding early because AI-assistant advertising behaves differently from everything before it. For the news timeline and specifics, see the report on image and video ads coming to ChatGPT.
This guide is about the surface, not the individual format. Whether the first rich unit is a video card or a conversational "ask the ad a question" experience matters less than the fact that a place people go to decide what to buy is becoming a place brands can pay to appear. That shift asks two concrete things of anyone who wants to show up there, and the back half of this guide is about both.
Keep the facts tight, because this is exactly the kind of story that gets exaggerated. Today, a ChatGPT ad is one standard unit — a headline, a short description, an image, and a link — that appears at the bottom of an answer when there is a relevant sponsored product or service, bought on a cost-per-click basis through a self-service manager. The job listings point to richer formats: image, video, native, conversational, and interactive. OpenAI has publicly floated a "conversational" unit where you could ask an ad questions before making a purchase decision.
Three constraints are already set. Ads run on separate systems from the chat model, so advertisers cannot shape, rank, or alter ChatGPT's answers. Conversations are kept private from advertisers. And ads appear only on the Free and Go tiers — ChatGPT Go is $8 a month — while Plus, Pro, Business, and Enterprise stay ad-free. Read those as the guardrails the ad product is being built inside, and treat any specific format claim as provisional until OpenAI ships it.
The instinct is to file this under "another ad platform." Resist it, because the two properties that make an AI-chat ad distinct are the whole point.
A feed ad interrupts someone who came to be entertained. A ChatGPT ad answers someone who just typed a question — often a question with commercial intent, like which product solves a problem they described in their own words. That is closer to search advertising than social, and it means the same ad creative that feels intrusive in a feed can feel useful under an answer. The context is richer than a keyword: the model already has the full framing of what the person is trying to do.
This is the part most coverage misses. Even before any ad loads, ChatGPT is already answering "what should I buy" questions organically — naming products, tools, and brands inside the response. So the surface has two layers of visibility: the paid unit at the bottom, and the organic mentions in the answer itself. You can pay for the first. The second you earn, the same way you earn a search ranking, by having a content footprint the model can find and cite. Ignoring that second layer to focus only on the ad slot is like buying search ads while never publishing a page.
Here is the strategic fork that separates this from the rest of 2026's ad news. Meta's Advantage+, TikTok's Symphony, and Snapchat's Ads Manager all moved to generate the ad creative for you — image-to-video, AI actors, auto-variants — pulling production inside the ad platform. OpenAI, at least in what it has signalled, is doing the opposite: building the ad placement and formats, and leaving the creative to the advertiser. If you want to run a video ad in ChatGPT, you bring the video.
That puts the production burden squarely back on your own pipeline. And it is a volume problem, not a one-asset problem — the moment a surface supports image and video, performance depends on testing many creatives, not shipping one. So the real question a marketer should ask about ChatGPT ads is not "how do I buy the placement" but "where do I get enough on-brand image and video to feed it, and the organic content to get cited above it." That question has a concrete answer, and it is the same answer for both jobs.
The surface asks for two things: visual ad creative you can run in the paid slot, and an organic content footprint that earns you a mention in the answer. A content engine is what produces both from the same source, which is why the production layer matters more here than the ad-buying mechanics.
This is a generation problem, and it is exactly what Kompozy is built for. Kompozy is an AI content engine — not a repurposing tool — that turns one input into finished assets across formats. On the image side that means Photo Posts, Infographic Photos, Persona Photos, Quote Graphics, and Carousel Posts; on the video side, Persona Shorts, Marketing Shorts, Clipped Shorts, and Listicle Video. Every asset is governed by one Persona Brief and a face-locked persona pool, so a whole test set of creatives stays recognizably yours instead of drifting. When a ChatGPT ad slot opens, you are not starting from a blank timeline — you already have on-brand image and video to run.
Being named in the answer is generative engine optimization, and it runs on the same fuel as search: consistent, on-brand content published where models can find it. Kompozy generates Text Posts, Blog Articles, and Email Newsletters alongside the visual formats, and publishes the whole set across nine social platforms plus blog and email from one scheduling queue. That breadth is what builds the footprint an AI model draws on when it decides which brands to mention. The same engine that fills your ad slot builds the presence that gets you cited above it — you are not running two separate content operations.
For the wider pattern of ad platforms absorbing creative production, see the guide on AI ad generation inside the ad platforms and the roundup of Meta AI multimedia ad best practices. ChatGPT is the inverse case — the platform that opened the surface but left the creative to you — which is precisely why the production engine, not the ad manager, is where the leverage sits.
You cannot buy a ChatGPT video ad today, and you should not wait to act as if you can. Two moves compound in the meantime. First, treat the news as a content moment and publish your own take across platforms while the topic is fresh — that is both organic reach and a signal to the models. Second, build the habit of generating on-brand image and video from every piece of source content you already make, so that when the surface opens you have a creative library and a citation footprint instead of a scramble. The advertisers who win the first ChatGPT ad auctions will be the ones who already had the content — the placement is new, but the work that fills it is the work you can start today.
OpenAI is building toward it. Engineering job listings reported on July 1, 2026 describe image, video, native, conversational, and interactive ad formats in development for ChatGPT, beyond the single text-and-image sponsored unit it tests today. No launch date has been announced, so treat it as a confirmed direction rather than a shipped feature.
Placement and intent. Social ads interrupt a feed you are scrolling; a ChatGPT ad appears at the bottom of an answer to a question you just asked, often a question with clear buying intent. OpenAI also says ads run on separate systems from the chat model and cannot alter the answer, and today they show only on the Free and Go tiers.
No. OpenAI is building the ad placement and formats inside ChatGPT; the advertiser still supplies the image and video. That is the opposite of Meta Advantage+, TikTok Symphony, and Snapchat's Ads Manager, which generate the creative for you. For ChatGPT ads you bring your own assets, which puts the production burden back on your content pipeline.
Generative engine optimization (GEO) is being surfaced and cited in AI-generated answers — the organic equivalent of ranking in search. As ChatGPT adds paid slots, the answer above the ad still recommends products, so a broad, on-brand content footprint that AI models can cite becomes a distribution channel that no ad buy replaces.
The same way you solve creative volume anywhere: generate it from one source instead of building each asset by hand. A content engine like Kompozy turns a single input into image formats (photo posts, infographics, quote graphics, carousels) and video formats (persona shorts, marketing shorts, clipped shorts, listicle video), all on-brand, so you have both ad creative and the organic footprint that earns citations.
OpenAI is building image, video, native, and conversational ad formats for ChatGPT — reported via July 2026 hiring — which would make an AI assistant a real ad surface for the first time. It differs from social advertising in placement and intent: the ad sits under an answer to a question the user just asked. Crucially, OpenAI builds the slot, not the creative, so brands still supply image and video. The two jobs it creates are producing that visual creative at volume and building an organic presence worth citing in the answer itself.
Get started → · ← All guides · Compare Kompozy vs other tools