OAI-SearchBot

When ChatGPT search answers a question with linked sources, those pages were discovered and indexed by OAI-SearchBot. It builds the search index that ChatGPT queries at answer time, making it the OpenAI bot with the most direct impact on whether you earn an AI citation. OpenAI explicitly separates it from GPTBot: OAI-SearchBot is for search, not training, so site owners can welcome search visibility while still opting out of training data collection.

Disallowing OAI-SearchBot in robots.txt is the AI-era equivalent of deindexing yourself from Google. Your pages drop out of ChatGPT's retrieval layer, so even perfectly optimized content cannot be surfaced or cited. Some sites blocked it accidentally with blanket wildcard rules written before AI crawlers existed, or by copying block lists that lump training and search crawlers together. Audit your robots.txt specifically: allow OAI-SearchBot even if you choose to block GPTBot, because the two decisions are independent and have opposite visibility consequences.

Treat OAI-SearchBot like a search engine crawler with limited patience. Serve fully rendered HTML, since it does not reliably execute JavaScript; keep key answers in crawlable text; and use clean structured data to clarify entities. Then verify it actually visits. Server-side bot tracking, such as Geonimo's AI traffic analytics, shows OAI-SearchBot crawl frequency per page, which is a leading indicator: pages it never fetches cannot appear in ChatGPT answers.