Is your site blocked from ChatGPT?
Most sites that blocked AI crawlers meant to opt out of training and took their citations down with it. This reads your robots.txt, checks all 20 tokens that matter, and shows you the exact line that decided each one.
Nothing is stored, and no email is asked for. Or try , , .
The two crawlers everyone confuses
GPTBotModel training
Collects pages that may be used to train OpenAI’s models. Blocking it is a real choice with a real rationale, and it costs you nothing in AI answers. If you do not want your writing in a training set, this is the token to disallow.
OAI-SearchBotCitations
Fetches pages so ChatGPT can quote them and link back. Blocking this removes you from ChatGPT’s answers entirely. Almost nobody blocks it deliberately; they block it by pasting a robots.txt snippet that happened to include it.
The same split runs through every vendor. Anthropic separates ClaudeBot from Claude-SearchBot. Google separates Google-Extended, which governs Gemini training, from Googlebot, which fills the index that AI Overviews and AI Mode are built from. Opting out of training and staying citable is a supported position at all three. It just takes naming the right token.
How often this goes wrong
We read the robots.txt of 170 well-known sites across 10 categories on 14 September 2026. 148 returned a file we could parse, and those are the denominator below.
News sites block; documentation and SaaS almost never do. Compare the first two figures: GPTBot is refused 2.3 times as often as OAI-SearchBot, so most sites that opted out of training kept their ChatGPT citations. The 12 sites blocking OAI-SearchBot are who this checker is for. Some chose it. Others pasted a list of user agents without knowing what each one governs, and the file looks the same either way. The full report is here.
What each of the 20 tokens does
Whether ChatGPT can surface and cite your pages
If you block it: ChatGPT search cannot cite you. This is the row that matters most, and the one most often blocked by accident.
Fetches made when a person asks ChatGPT to open your link
If you block it: Someone who pastes your URL into ChatGPT gets an error instead of your page.
Whether your pages can be used to train OpenAI models
If you block it: Nothing, for citations. Blocking this is a legitimate choice and does not remove you from ChatGPT search.
Whether Claude can surface and cite your pages
If you block it: Claude's search cannot cite you.
Fetches made when a person asks Claude to open your link
If you block it: Someone who pastes your URL into Claude gets an error instead of your page.
Whether your pages can be used to train Anthropic models
If you block it: Nothing, for citations. Blocking this does not remove you from Claude's search results.
Google Search, and with it AI Overviews and AI Mode
If you block it: Everything. AI Overviews and AI Mode are built from the Google index, so blocking Googlebot removes you from both as well as from Search.
Gemini model training and grounding
If you block it: Not your ranking, and not AI Overviews. This token is widely believed to control AI Overviews and does not: those come from the ordinary index.
Whether Perplexity can index and cite your pages
If you block it: Perplexity cannot cite you.
Fetches made when a person follows your link inside Perplexity
If you block it: A user who clicks through to you from Perplexity may not reach the page.
The Bing index, which is what Microsoft Copilot answers from
If you block it: Copilot has nothing of yours to answer with, because it answers from Bing.
Siri, Spotlight and Safari suggestions
If you block it: Apple's search surfaces have nothing of yours to show. This is a search crawler, not only an AI one.
Whether your pages train Apple's foundation models
If you block it: Nothing you can see. Apple states this governs training only, and disallowing it does not stop Applebot crawling you.
Search experiences across Amazon products and services
If you block it: Amazon's search surfaces cannot surface you.
Fetches behind user actions, such as an Alexa question needing current information
If you block it: Someone asking Alexa a question your page answers gets an answer from somewhere else.
Improving Amazon's products and services
If you block it: Nothing for the search surfaces, which Amzn-SearchBot governs separately.
DuckDuckGo's AI-assisted answers, crawled in real time
If you block it: DuckAssist cannot quote you. DuckDuckGo states this does not affect your organic rankings either way.
The index Mistral's assistant answers from
If you block it: Mistral has nothing of yours to retrieve.
Fetches made when a person asks Mistral to open your link
If you block it: A reader who follows your link inside Mistral may not reach the page.
The public Common Crawl dataset, which many model builders train on
If you block it: Nothing directly, and a great deal indirectly: this is one refusal that reaches every model trained on the corpus.
Each token links to the vendor’s own documentation, which is the only place these are authoritative. DeepSeek is not listed because it publishes no crawler token to check.
A page on each one: user agent, IP ranges, how many sites block itThen the other half: being worth quoting
Access is the precondition, not the outcome. One guide per engine, each with what it actually takes once the crawler is in.
OAI-SearchBotClaude-SearchBotGooglebotPerplexityBotbingbotno published tokenQuestions
- Is my site blocked from ChatGPT?
- Put your domain in the box above and it will tell you. The token that decides is OAI-SearchBot: if your robots.txt disallows it, ChatGPT's search cannot quote your pages or link to you. GPTBot is a different crawler that governs model training only, and blocking it does not remove you from ChatGPT search.
- What is the difference between GPTBot and OAI-SearchBot?
- GPTBot collects pages that may be used to train OpenAI's models. OAI-SearchBot fetches pages so ChatGPT can cite them in an answer. They are separate tokens with separate jobs, and most sites that mean to opt out of training want to block the first and keep the second. Blocking both is the common mistake, usually made by copying a robots.txt snippet from a blog post.
- Does blocking Google-Extended remove me from AI Overviews?
- No, and this catches a lot of people. Google-Extended governs Gemini training and grounding. AI Overviews and AI Mode are built from the ordinary Google index, which Googlebot fills. Blocking Google-Extended leaves you in AI Overviews; blocking Googlebot removes you from AI Overviews, AI Mode and Search together.
- Why does the checker say a crawler is allowed when it is not getting in?
- Because robots.txt is one gate and not the only one. A site can also refuse a crawler at the CDN, by IP range, or with a bot challenge, and none of that is written in robots.txt. This tool reports what your robots.txt permits, which is the part you control from your own server. If robots.txt says yes and a crawler still is not reaching you, the block is at the edge.
- Will allowing these crawlers get me cited?
- It makes it possible, which is not the same as likely. Access is the precondition: a crawler that cannot read you cannot quote you, so a block is a hard ceiling. Everything above that ceiling is whether the page answers a question clearly enough to be worth quoting.
- Is it safe to let AI crawlers in?
- Allowing a retrieval crawler means AI answers can quote a sentence or two of your page and link back to you, which is a referral. Allowing a training crawler means your text may end up in a model. Those are genuinely different decisions and reasonable sites make them differently, which is why the results here separate them instead of scoring you out of ten.
Getting in is the easy half
This checks whether crawlers are allowed. The free scan reads your real pages and runs 42 checks on what they find once they are in.