Definition GPTBot collects page content to train ChatGPT models. It identifies as User-Agent: GPTBot. The OAI-SearchBot used for citation retrieval is a separate crawler.
How it works In its crawl cycle, GPTBot fetches robots.txt and skips all paths if disallowed. Allowed pages are pulled via normal HTTP requests. Fetched content enters OpenAI's data pipeline and may be used as training data in later model versions. The bot publishes IP ranges and uses reverse DNS for verification.
Where you see it in Scope Trends **Crawl Access** report detects your GPTBot rule in robots.txt. The **Bot Preference** setting applies allow or block with one click. **Server Logs** shows GPTBot requests at the daily level.
Frequently asked questions **If I block GPTBot, do I disappear from ChatGPT?** You exit model training. But ChatGPT Search citations rest on OAI-SearchBot, which is a separate bot.
**Does blocking GPTBot affect SEO?** No effect on classic Google SEO rankings. Only the OpenAI training set is impacted.