MetricRaven

What is Google-Safety?

Google · Other bots · User agent + IP range

Google's abuse crawler, for example to discover malware behind links posted publicly on Google properties.

A visit usually means: A product is checking the page for a specific job: an ad review, a link preview, a feed or an SEO index.

Google-Safety user agent

The full user agent string, as Google publishes it:

  • User-Agent
    Google-Safety

Version numbers change, so match the token Google-Safety anywhere in the user agent, ignoring case. Source: Google's crawler documentation, checked Nov 18, 2026.

Reading the string

  • It has no browser parts (no Mozilla/5.0 prefix), so it's easy to tell apart from visitors.

Paste a string from your logs into the user agent parser to see which crawler it is. New to the term? Read what a user agent is.

Block or allow Google-Safety in robots.txt

Google says Google-Safety doesn't follow robots.txt. Block it at your server or firewall if you need to.

Block everywhere
robots.txt
User-agent: Google-Safety
Disallow: /
Keep out of a folder
robots.txt
User-agent: Google-Safety
Disallow: /private/
Allow
robots.txt
User-agent: Google-Safety
Allow: /

A crawler follows the most specific group that names it, so a rule here overrides your User-agent: * group for this crawler. Well-behaved crawlers re-read robots.txt within about a day.

Google-Safety and Google's other crawlers

Google runs 17 crawlers in this directory, each with its own robots.txt token and its own job. What blocking each one costs you:

  • Google-Safety · OtherThe Google feature it checks pages for; it sends no visitors
  • Google-Agent · AnswersGoogle's assistant reading your pages when someone asks about them
  • Google-GeminiNotebook · AnswersGoogle's assistant reading your pages when someone asks about them
  • Google-NotebookLM · AnswersGoogle's assistant reading your pages when someone asks about them
  • Google-Read-Aloud · AnswersGoogle's assistant reading your pages when someone asks about them
  • Gemini-Deep-Research · AnswersGoogle's assistant reading your pages when someone asks about them
  • Google-CloudVertexBot · IndexingYour pages in Google's AI search answers
  • Google-Extended · TrainingModel training and datasets; it sends no visitors
  • GoogleOther-Image · TrainingModel training and datasets; it sends no visitors
  • GoogleOther-Video · TrainingModel training and datasets; it sends no visitors
  • GoogleOther · TrainingModel training and datasets; it sends no visitors
  • Googlebot · IndexingYour pages in Google's search results
  • Google-InspectionTool · IndexingYour pages in Google's search results
  • Storebot-Google · IndexingYour pages in Google's search results
  • AdsBot-Google · OtherThe Google feature it checks pages for; it sends no visitors
  • Mediapartners-Google · OtherThe Google feature it checks pages for; it sends no visitors
  • FeedFetcher-Google · OtherThe Google feature it checks pages for; it sends no visitors

Opt out of Google's AI training, stay in its search

Name only the training tokens. Google-CloudVertexBot, Googlebot, Google-InspectionTool, Storebot-Google have their own tokens, so they keep following your other rules.

robots.txt
User-agent: Google-Extended
Disallow: /

User-agent: GoogleOther-Image
Disallow: /

User-agent: GoogleOther-Video
Disallow: /

User-agent: GoogleOther
Disallow: /

Google-Safety IP ranges and verification

Verified when the user agent and the IP range both match

  1. 1User agent

    The request's user agent contains Google-Safety.

  2. 2Request IP

    Your server-side tracker sends the IP that made the request.

  3. 3Range check

    The IP is compared with the ranges Google publishes, refreshed twice a day.

Range sources

Check by reverse DNS

Google documents a DNS check you can run on any IP in your logs: a reverse DNS lookup of the request's IP should return a hostname ending in google.com, and a forward lookup of that hostname should return the same IP.

terminal
host <ip>          # → a name ending in google.com
host <that name>   # → the same <ip>

What Google-Safety means for your traffic and revenue

Google-Safety checks pages for one product job, such as an ad review or a feed. It isn't a visitor and shouldn't be counted as one.

Most crawlers never run JavaScript, so browser analytics can't see them, and the ones that do shouldn't count as visitors. MetricRaven's tracker ignores Google-Safety, and server-side crawl tracking shows which pages it reads instead. See Bots & AI crawlers.

Questions

What is Google-Safety?
Google's abuse crawler, for example to discover malware behind links posted publicly on Google properties. MetricRaven groups it under Other bots: ad checks, link previews, feed readers, SEO tools and other bots.
What is the Google-Safety user agent?
Google documents this user agent: "Google-Safety". Match the token "Google-Safety" rather than the whole string, because version numbers change.
How do I block Google-Safety in robots.txt?
Add "User-agent: Google-Safety" followed by "Disallow: /" to your robots.txt. Google says Google-Safety doesn't follow robots.txt. Block it at your server or firewall if you need to.
Does Google-Safety publish IP ranges?
Yes. Google publishes the addresses it sends Google-Safety from: https://developers.google.com/static/crawling/ipranges/special-crawlers.json. A request that claims to be Google-Safety from any other IP is an impostor. You can also check DNS: a reverse DNS lookup of the request's IP should return a hostname ending in google.com, and a forward lookup of that hostname should return the same IP.
Is Google-Safety safe?
The real Google-Safety only requests public pages, and Google documents it. The risk is impostors using its name, which the published IP ranges catch. Whether to allow it depends on what you get back: a product is checking the page for a specific job: an ad review, a link preview, a feed or an SEO index.
What other crawlers does Google run?
Google-Agent (AI assistant), Google-GeminiNotebook (AI assistant), Google-NotebookLM (AI assistant), Google-Read-Aloud (AI assistant), Gemini-Deep-Research (AI assistant), Google-CloudVertexBot (AI search), Google-Extended (AI training), GoogleOther-Image (AI training), GoogleOther-Video (AI training), GoogleOther (AI training), Googlebot (search engine), Google-InspectionTool (search engine), Storebot-Google (search engine), AdsBot-Google (other), Mediapartners-Google (other) and FeedFetcher-Google (other). Each has its own robots.txt token, and each does a different job.
Can I block Google's AI training but stay in its search?
Yes. Add a "Disallow: /" group for Google-Extended, GoogleOther-Image, GoogleOther-Video and GoogleOther. Google-CloudVertexBot, Googlebot, Google-InspectionTool and Storebot-Google have their own tokens, so they keep following your other rules.

See when AI assistants and crawlers read your site

Add one line to your server and MetricRaven shows which assistants, search engines and training crawlers request your pages, which ones are IP-verified, and the URLs they expected to find, including 404s.