Overview
GPTBot is associated with OpenAI and used for model-training.
The primary identifying signal is GPTBot.
Representative request identity: GPTBot/1.4.
Directory status: Active. Identity classification: Verified with High confidence. Identity type: official-documented.
This identity is recorded as honoring robots.txt under `GPTBot`.
Common use: crawling public web content that may be used to improve and train OpenAI foundation models.
Risk classification: Neutral. Recommended handling: depends.
Identity
- User-Agent Pattern
-
GPTBot - Aliases
- OpenAI training crawler
- HTTP Agent Examples
-
GPTBot/1.4 - Robots Token
- GPTBot
- Identity Type
- Officially documented
- Evidence Method
- Match `GPTBot` as the documented identity signal and corroborate source network, DNS, signatures, or request context when stronger authentication is required.
Classification
- Type
- AI
- Kind
- Crawler
- Family
- OpenAI
- Purpose
- Model training
Behavior and handling
- Common Use
- crawling public web content that may be used to improve and train OpenAI foundation models
- Detection Notes
- Match the `GPTBot` identity in request logs and corroborate the surrounding request context. A User-Agent string alone can be copied.
- Respects robots.txt
- Yes
- Spoofing Risk
- The `GPTBot` signal can be copied by unrelated clients. Treat it as a claimed identity unless corroborated by stronger request or network evidence.
- Risk
- Neutral
- Recommended Handling
- Depends
Rules and controls
- Robots.txt Snippet
-
User-agent: GPTBot Disallow: /
Relationships
- Operator
- OpenAI Checked 2026-10-05
Relationships without an Evidence link are normalized from the canonical directory record. They should not be interpreted as independent proof of physical presence or request origin.