GPTBot
What is GPTBot?
About
GPTBot is a web crawler operated by OpenAI to download training data for its LLMs (Large Language Models) that power AI products like ChatGPT.
Detail
Operator | OpenAI |
Documentation | https://platform.openai.com/docs/bots |
Type
Expected Behavior
It's generally unclear how AI data scrapers choose which websites to crawl and how often to crawl them. They might choose to visits websites with a higher information density more frequently, depending on the type of AI models they're training. For example, it would make sense that an agent training an LLM (Large Language Model) would favor sites with a lot of regularly updating text content.
Insights
GPTBot's Activity on Your Website
Other Websites
Access Control
Should I Block GPTBot?
It's up to you. AI data scrapers usually download publicly available internet content, which is freely accessible by default. However, you might want to block them if you're concerned about attribution or how your creative work could be used in the resulting AI model.
Using Robots.txt
User Agent Token | Description |
---|---|
GPTBot |
Should match instances of GPTBot |
You can block GPTBot or limit its access by setting user agent token rules in your website's robots.txt.
# robots.txt
# This should block GPTBot
User-agent: GPTBot
Disallow: /