GPTBot is OpenAI's web crawler that gathers training data for future versions of GPT models. It is one of three distinct OpenAI bots — the other two being OAI-SearchBot (which builds OpenAI's own search index for ChatGPT Search) and ChatGPT-User (the on-demand fetcher that fires when a user pastes a URL into ChatGPT).
Allowing GPTBot in robots.txt makes your content eligible for inclusion in future model training. It does not directly affect citation rates in current ChatGPT or ChatGPT Search — those depend on OAI-SearchBot and the Bing index. Blocking GPTBot has zero impact on classic Google rank but excludes you from contributing to future GPT model knowledge.
The user-agent string is: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.0; +https://openai.com/gptbot). To allow it explicitly, add a "User-agent: GPTBot" stanza with "Allow: /" to your robots.txt.
For most sites the high-leverage decision is OAI-SearchBot and ChatGPT-User accessibility (these power citations now). GPTBot is the slow lane — it influences how future model versions know about your content, which only matters when OpenAI ships a new model. Both choices are independent: you can allow or block each bot separately.