ChatGPT Ads Trust Layer

In conversational AI, a sponsored route sits close to advice. The trust layer must therefore be explicit before the offer is surfaced: sponsor identity, provenance, disclosure, negative contexts, and non-behavioral matching.

DisclosureSponsored surfaces carry explicit labels and separation rules.

ProvenanceOffer metadata declares sponsor domain, verification method, and trust lineage.

PrivacyMatching is based on current context rather than cross-site behavioral profiles.

RoutingAgents inspect agent_ad objects before referral or action.

Machine-readable entry points

agent-ad.json defines the sponsored offer object. conversational-ad-context.json separates the buyer situation, current-conversation hint, eligibility, disclosure, landing-page evidence, negative contexts, and aggregate measurement. trust-resolver.json declares attribution and licensing. agent-discovery.json exposes the broader discovery map.

Operational boundary

ChatGPT Ads' European expansion does not erase the separation between answer and ad. It makes that separation measurable. A trustworthy implementation must preserve an independently generated answer, label the adjacent commercial route, avoid sensitive contexts, and keep advertiser reporting aggregate and non-identifying.

Scope

This is compatibility language for the answer-engine advertising paradigm. It does not claim a platform integration.