OpenAI’s new ad system will target you inside ChatGPT using your own conversations

ChatGPT’s advertising system uses conversational context to decide which sponsored placement may be relevant beneath a chat. Official product documentation says the selection can draw on the intent of the current conversation and, when personalization is enabled, selected signals from a broader ChatGPT experience.

Advertisers do not receive the private conversation itself, and the ad is presented below rather than inside the model’s answer. Those boundaries matter, but they do not change the core targeting fact stated in the supplied title.

Current conversation context enters ad selection

OpenAI’s official ads help page says delivery considers the context and intent of the current conversation along with the landing page, creative, and advertiser-provided context hints.

That directly supports conversation-based targeting within ChatGPT. For current conversation context enters ad selection, OpenAI’s official ads help page is the controlling account because it supplies the named event, object, or measurement rather than a detached retelling. Its 2026-07-31 date also fixes the evidence to the article’s verification window. Reading the source at that level keeps the central claim in its documented setting and blocks a dramatic detail from expanding into conditions the source never examined.

Broader account signals require personalization

The same product documentation adds that selected signals from a user’s broader ChatGPT experience may be considered when ads personalization is enabled.

The setting distinction prevents the claim from implying identical history use for every account. This section adds a distinct layer to the record: broader account signals require personalization explains how the central observation should be interpreted, not merely repeated. The attribution to The same product documentation matters because method, location, and scope travel with the fact. Those limits are especially important for a claim classified as current official product policy; removing them would make a supported detail sound broader or more current than the evidence allows.

Ads appear below the answer

OpenAI’s advertiser site describes sponsored placements built for conversational moments while the product help page places the unit below relevant conversations.

The documented format keeps the ad visually separate from the assistant’s substantive response. The source behind ads appear below the answer does more than provide another citation. OpenAI’s advertiser site establishes the comparison or mechanism used in this section and shows which part remains inference. That separation is why the paragraph can preserve a vivid description without converting analogy into measurement, possibility into certainty, or one documented case into a rule for every similar event.

Sensitive conversations are excluded

OpenAI’s July 15 ad policy bars placements near sensitive, vulnerable, inappropriate, and brand-unsafe conversation categories.

The restrictions show that conversational relevance is bounded by safety and trust rules. In the evidence chain, sensitive conversations are excluded connects the observable record with its practical or historical consequence. OpenAI’s July 15 ad policy controls that connection, so the article keeps the same subject and factual state instead of importing a more dramatic conclusion. The result is topic-specific context that remains auditable: the claim can be traced to a source, date, and defined setting rather than to an unsupported generalization.

Eligibility excludes several paid plans and minors

The official eligibility section says ads are not shown to Plus, Pro, or Business users or to users identified as under 18.

The targeting system therefore applies only to eligible accounts within the test or rollout. The significance of eligibility excludes several paid plans and minors is therefore also a limit on the story. The official eligibility section supports the stated conclusion as of 2026-08-03, while leaving later updates or unresolved interpretation visible. Holding that boundary through the final section keeps the body consistent with the supplied title and makes clear which future evidence could revise the account without rewriting what the cited record currently establishes.

The official documentation supports the essential claim: conversation content and intent can influence which ad appears, with broader experience signals available when personalization is on. The same sources also establish limits—no ad inside the answer, exclusions for sensitive contexts, and eligibility rules. Those limits define the system rather than contradicting its conversational targeting.

Taken together, current conversation context enters ad selection and eligibility excludes several paid plans and minors define the article’s evidentiary range. The intervening sections on broader account signals require personalization, ads appear below the answer, and sensitive conversations are excluded show how the record moves from observation to context without changing factual state. That sequence is why the current official product policy classification remains appropriate as of 2026-08-03: each major point has a named source, while uncertainty stays attached to the question it actually affects.

For this Science record, the source trail runs from OpenAI’s official ads help page to The official eligibility section. That combination supports the event or evidence date recorded as 2026-07-31 and prevents a source with a different purpose from controlling the conclusion. The verification is consequently reproducible: another editor can open the same cited pages, check the same title-level assertion, and see why the article preserves its stated scope.

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

More from Morning Overview