Brands buying ChatGPT Ads still cannot see where their advertisements actually appeared inside the chatbot's answers. That single gap sits at the centre of a growing argument between advertisers and OpenAI, and it is the question the trade is now asking out loud: if you cannot audit the placement, what exactly did you pay for?

Byline: Sophie Lindqvist.

What does this change for advertisers?

The answer depends on what a brand thinks it is buying. In search and social, an ad placement comes with a report: the page, the query, the context, and often a domain-level list of where the creative served. That report is the basis of brand-safety checks, fraud detection and negotiation over price.

ChatGPT Ads, by contrast, do not hand over an equivalent receipt. The advertisement surfaces inside a generated answer, stitched into a conversation rather than anchored to a URL. The advertiser sees that the ad ran and can measure outcomes such as clicks. What they do not see is the surrounding text, the specific conversation, or the placement mechanics that put their brand next to a given response.

This matters because placement context is not decoration. It is the thing buyers use to decide whether a channel deserves budget at all.

Why the ad world is sceptical

Ad buyers have been here before. Programmatic advertising spent a decade fighting the same problem at scale: inventory resold through chains of intermediaries, ads landing on pages nobody checked, and brands pulling spend after audits exposed the gap.

The industry's answer was verification — third parties that log where an ad appeared and score the context. That solution depends on one thing: an addressable, inspectable location for every impression. A chat interface removes the address. There is no stable page, no permanent link, and no way to replay the exact generated answer after the fact.

So the usual verification tools do not simply need tuning. They need a new mechanism, and nobody has shipped one yet.

Is this unique to OpenAI?

Partly. Any conversational assistant that runs ads inside generated text faces the same structural problem, and competitors moving into AI-placed advertising will inherit it. The difference with OpenAI is speed and reach: ChatGPT carries enormous daily traffic, and advertisers feel pressure to test the channel early rather than arrive late.

That creates an uncomfortable asymmetry. The platform knows exactly which conversations triggered which ads. The buyer knows only what the platform chooses to report. Every intermediary problem the programmatic era exposed — opaque inventory, unverifiable context, trust me as a business model — reappears here in a new wrapper.

What happens next?

Expect the argument to follow a familiar script. Advertisers will demand placement-level reporting, even in aggregate form: what topics the ads appeared against, what kinds of prompts triggered them, and how often answers mentioning a competitor sat next to a brand's creative.

OpenAI has commercial reasons to move carefully. Publishing placement detail could expose how its ad selection works, and how often answers containing errors or contested claims carried paid messages. Transparency cuts both ways.

For now, buyers face a plain choice. They can treat ChatGPT Ads as an experimental line item — small budgets, strict outcome metrics, and no assumptions about context — or they can wait until the platform offers the receipts every other ad channel eventually had to produce.

The question in the headline is not rhetorical. Until placement visibility arrives, the honest answer is: brands are buying outcomes and a promise, and nothing they can independently check.