OpenAI details ChatGPT ad stack built around trust and relevance

In its engineering disclosure published on 21 October 2026, OpenAI said ChatGPT Ads uses stateful sessions and a post-trained LLM orchestrator to understand the conversation and turn it into queries and embeddings. The system then retrieves candidates through approximate-nearest-neighbor indices and calls tools for eligibility, relevance and engagement scoring. OpenAI said filters address targeting, policy, budgets, pacing and deduplication, while the orchestrator nominates candidates and an auction tool, not the LLM itself, determines final selection and pricing. The October disclosure said measurement and optimization rely on Pixel and Conversions API foundations. OpenAI described the implementation as a first-party engineering account rather than an independent audit, and the material does not disclose external verification of the system's performance or user impact.

BeCrypto News shows a short excerpt and links to the publisher. The full report belongs to Blockchain.News.