# Google-Extended

> A robots.txt user-agent token that lets a site opt out of having its content used to train and ground Google's Gemini models, while leaving Google Search indexing unaffected.

_The Agentic Web Lexicon · /glossary/google-extended · [JSON](/api/glossary/google-extended) · [all The Agentic Web Lexicon](/glossary)_

- **term:** Google-Extended
- **category:** identity
- **short_def:** A robots.txt user-agent token that lets a site opt out of having its content used to train and ground Google's Gemini models, while leaving Google Search indexing unaffected.
- **long_def:** Google-Extended is a control token, not a crawler. Adding 'User-agent: Google-Extended' with 'Disallow: /' to robots.txt tells Google not to use the site's content for training or grounding Gemini and Vertex AI generative models; normal Googlebot search crawling continues. Introduced by Google in September 2023.
- **see_also:** —
- **etymology_origin:** Introduced by Google in September 2023 as an opt-out control for generative-AI training/grounding.
- **related_to:** —
- **contrast_with:** Unlike a crawler token that blocks fetching, Google-Extended only governs AI training/grounding use — Googlebot search access is unaffected.
- **example:** A publisher adds Google-Extended to robots.txt to keep its articles out of Gemini training while staying in Google Search.
- **source:** https://developers.google.com/search/docs/crawling-indexing/overview-google-crawlers
- **status:** stable
- **why_it_matters:** It separates 'be findable in search' from 'be used for AI training' — a core opt-out lever in the access-economics debate.
- **sameAs:** —
- **bridge_entity:** /access-economics
- **last_verified:** 2026-06-15
- **md_twin:** /glossary/google-extended.md