ChatGPT ads can use the context of your current chat

ChatGPT ads can use the context and intent of a person’s current chat to decide which sponsored placement may be relevant, according to OpenAI’s product documentation. That is a more precise statement than saying an advertiser gets to read a conversation or shape the answer ChatGPT gives.

OpenAI says ads are labeled and visually separate from responses, and that advertisers do not receive users’ chats, chat histories, memories or personal details. The distinction is important because relevance systems can use information inside a service without handing the underlying conversation to the company that bought the ad.

Current-chat context is one signal in the ad system

OpenAI’s ads help article says its system considers the context and intent of the current conversation, alongside an ad’s landing page, title, copy and advertiser-provided context hints. If a placement is eligible, it can appear below the end of a response rather than inside that response.

Those details set practical limits on how relevance is described. Not every question produces an ad, and a placement is not based solely on a single prompt. The current chat can be a signal. OpenAI also says longer conversations may be considered with overall context and user experience in mind, which means relevance is a system decision rather than a direct keyword match.

Personalized ads add settings-controlled signals

With ad personalization switched off, OpenAI says a user may still see ads based on the current chat thread, but the company will not use other chats, ad history or topics to inform those ads. With personalization enabled, it says select signals from broader ChatGPT experience may also be used, including past chats and memory where those features are on.

That setting is separate from the basic fact that a conversation has a topic. The help material says users can manage ad controls, see why an ad was shown, clear data used for ads and turn off personalization. OpenAI’s Memory FAQ separately explains that memory controls can be managed and that Temporary Chat is designed not to use or update memory.

Advertisers do not get to alter ChatGPT’s answer

OpenAI’s documentation states that ads run on separate systems from the chat model and that advertisers cannot shape, rank or alter ChatGPT responses. Sponsored items are paid placements, not endorsements. That boundary is useful for readers evaluating a recommendation: an ad beside an answer is not itself a fact claim made by the assistant.

The company also says advertisers receive aggregated, non-identifying reporting such as views and clicks in the early test. Its privacy policy and advertising documentation should be read together, because product settings and availability can change over time. A person who messages an advertiser through an ad is making a separate, direct communication; OpenAI says the advertiser sees those direct messages, not the person’s ordinary chats.

Some chats and users are outside the test experience

OpenAI says ads are not eligible near sensitive or regulated conversations, including personal health, mental health and politics. It also says temporary chats do not show ads and that accounts identified as belonging to people under 18 are not shown ads. Availability varies by region and, according to the help article, ad controls are limited to Free and Go users in regions where ads are available or coming soon.

For someone trying to understand a placement, the practical question is not simply whether an ad appeared after a certain prompt. The relevant questions are whether the account is in an eligible region, whether the conversation is eligible, and whether ad personalization is enabled. The in-product explanation for a particular ad and the current ad settings are more useful than assumptions based on another person’s screenshot or account type.

It is also sensible to distinguish an advertisement from the service’s ordinary answer. A sponsored card is identified as an ad and is separate from the response; following it takes a person to an advertiser’s own destination. Before sharing information or buying anything, a reader can review the advertiser, the destination and the terms of the offer just as they would for any other online advertisement. That is ordinary consumer caution, not evidence that the chat itself was disclosed to the advertiser.

The company’s ad policies set additional limits on what advertisers may promote and prohibit ads around vulnerable user-model interactions. Those rules should not be confused with a promise that no ad will ever be imperfect; they are the stated standards governing an early program. Users who see a questionable placement can dismiss it or submit feedback through the ad controls.

The most accurate summary is therefore straightforward. ChatGPT ads can use the context of a current chat, and broader signals can be used only when personalization is enabled. Advertisers do not receive the underlying conversations, and they cannot change ChatGPT’s answers. OpenAI’s advertising terms make clear that the program and its delivery can evolve, so current in-product settings remain the best check for an individual account.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview