--- language: en title: "GPTBot" description: "GPTBot is OpenAI’s crawler for content that may contribute to model training. Learn how to distinguish this collection from access for ChatGPT search." canonical: "https://tropk.ai/en/resources/glossary/gptbot" last_modified: "2026-10-12T16:11:32.785Z" --- # GPTBot Reviewed on 2026-10-06. ## What it means GPTBot is OpenAI’s [crawler](/en/resources/glossary/crawler) that collects web content that may be used to train its [generative AI](/en/resources/glossary/generative-ai) foundation models. The documentation distinguishes this purpose from [ChatGPT](/en/resources/glossary/chatgpt) search. The site owner can use a specific [robots.txt](/en/resources/glossary/robots-txt) rule to indicate that the site’s content should not be used for this training. ## Why it matters for marketing For marketing, the decision involves permitted use of the content library, especially when it includes material with restrictive licenses, [original research](/en/resources/glossary/original-research), or partner content. Blocking all agents from a company for a single purpose may conflict with editorial intent. The team needs to separate its training policy from page availability for search discovery. ## Example in practice Imagine a publisher with public manuals whose license does not allow use in training. It decides to restrict GPTBot but keep [OAI-SearchBot](/en/resources/glossary/oai-searchbot) permitted to discover the manuals. The evidence to monitor is the rule served in robots.txt and the recorded behavior of requests. This choice establishes different permissions; it does not demonstrate the exclusion of material already collected. ## How to apply or verify The person responsible for infrastructure should check the robots.txt file actually served by the domain, including rules specific to GPTBot, and examine access logs after the change. To attribute visits, compare the declared agent with the IP ranges published by OpenAI. Record the date, path, and server response; a name in the header can be imitated. ## Important distinctions GPTBot is not OAI-SearchBot or [ChatGPT-User](/en/resources/glossary/chatgpt-user): training, search, and user actions have distinct controls and purposes. When both crawlers are permitted, OpenAI says it may use a collection for both purposes. Restricting GPTBot does not prove retroactive removal from training, nor does it, by itself, guarantee or prevent a [citation](/en/resources/glossary/ai-citation) in a response. ## Sources and references - [Overview of OpenAI Crawlers]() — OpenAI ## Continue a leitura - [Training data]() - [OAI-SearchBot]() - [robots.txt]() ## About the author [Vinicius Pacheco]() Responsible for Tropk’s technical content, focusing on crawling, indexing, structured data, and agent access to content. --- [All pages in Markdown]()