Is My Website in ChatGPT Training Data? What You Can Check and What You Can't

There is no official way to look up whether your website is in ChatGPT training data: OpenAI does not publish a list of trained sites or offer a lookup tool. What you can check is the evidence around it, namely whether GPTBot has crawled you and whether your pages are in Common Crawl, a public web archive that OpenAI used for GPT-3. This guide shows both checks step by step and explains what blocking does and does not change.

Published: 2026-10-23 · 7 min read · Satelo

Training and search are two different paths

People often mix up two questions: "did ChatGPT learn from my site?" and "can ChatGPT find my site?". They are separate systems with separate crawlers.

OpenAI's crawler overview describes GPTBot as the bot that crawls content that "may be used" for training foundation models, and says robots.txt settings for each bot are independent. Blocking GPTBot does not remove you from ChatGPT search. If search is your actual goal, our guide to getting indexed by ChatGPT is the right starting point.

What OpenAI actually discloses about training data

OpenAI's help article How ChatGPT and our foundation models are developed is the main official source. It says:

• Models are developed from three kinds of information: publicly available internet content, information accessed through third-party partners, and information provided or generated by users, human trainers and researchers. It also mentions growing use of synthetic data.
• For public web content, OpenAI says it uses only freely accessible information, does not intentionally gather data from sources known to be behind paywalls or from the dark web, and filters out material such as hate speech, adult content, sites that aggregate personal information, and spam.
• Website owners can manage whether public content is accessed for training "by using standard web controls such as robots.txt to disallow GPTBot".
• Models "do not store or retain copies of the data they are trained on"; training adjusts parameters instead.

What the article does not say: which sites were used, which crawls fed which model, or whether a given URL made it through the filters. We found no OpenAI page that answers those questions for an individual site, so anyone who claims to "look up your site in ChatGPT's training set" is guessing.

Where Common Crawl fits in

Common Crawl is a nonprofit that publishes a free archive of the web, collected by its crawler CCBot. It matters here because OpenAI documented using it. The GPT-3 paper, Language Models are Few-Shot Learners (2020), lists "Common Crawl (filtered)" as 60% of the training mix and says the data came from 41 monthly Common Crawl shards covering 2016 to 2019, filtered for quality and deduplicated.

Two cautions before you draw conclusions:

• That is a statement about GPT-3. OpenAI's current help article names categories of data, not specific datasets, so we cannot say whether or how Common Crawl is used for current ChatGPT models.
• Even for GPT-3, being in Common Crawl did not guarantee inclusion: the paper describes filtering by similarity to high-quality reference corpora.

So presence in Common Crawl is a useful signal that your pages are in the most widely used public web archive, which many AI developers draw on. It is not proof that ChatGPT trained on them.

Check 1: is your site in Common Crawl?

Common Crawl runs a free index server at index.commoncrawl.org. Each monthly crawl has its own index, listed on that page with its date range.

1. Open the index page and pick a recent crawl (the IDs look like CC-MAIN-YYYY-WW).
2. Query it for your domain. In a browser or terminal: https://index.commoncrawl.org/CC-MAIN-YYYY-WW-index?url=yourdomain.com/*&output=json
3. Read the result. Each line is one captured page with its URL, timestamp and the HTTP status the crawler got. No lines (or a "No Captures found" message) means that crawl did not capture your domain.
4. Repeat for two or three older crawls. Coverage varies a lot between months, especially for small sites.

Common Crawl's FAQ warns that this CDX API is heavily rate limited because it is frequently abused: send one query at a time, pause between calls and avoid parallel scripts. For large-scale checks it suggests querying the URL index through Amazon Athena or Spark instead.

What to look at in the results: the status field. Captures with 200 contain your content. Captures with 301, 403 or 404 mean the crawler visited but got no page, which often points to redirects or bot protection.

Check 2: has GPTBot crawled you?

Your server access logs are the only direct evidence of OpenAI's training crawler visiting.

1. Download raw access logs from your host (often under "Logs" or "Raw access" in the control panel) or read them on the server.
2. Search for the user agent token: grep "GPTBot" access.log. OpenAI's documented user agent string contains GPTBot/1.4; +https://openai.com/gptbot.
3. Verify the IP addresses. User agents are trivial to fake. OpenAI publishes GPTBot's IP ranges in gptbot.json; a request claiming to be GPTBot from an IP outside those ranges is not OpenAI.
4. Note the status codes and which URLs were fetched. 200 responses mean the bot received the content.

Logs only go back as far as your host keeps them, often weeks, so they cannot tell you about crawling in earlier years. And a GPTBot visit means content "may be used" for training, not that it was.

Check 3: asking ChatGPT (a weak signal)

You can ask ChatGPT, with web search turned off, what it knows about your site or brand. If it describes your specific content accurately, that suggests the information was somewhere in its training data, possibly from other sites that mention you. Treat this as the weakest of the three checks: models can produce confident but wrong descriptions, and a blank answer does not prove absence. OpenAI's help article itself notes that the same question may yield different answers across queries.

What blocking GPTBot and CCBot does (and does not) do

Both crawlers follow robots.txt. OpenAI says disallowing GPTBot signals that your content should not be used for training. Common Crawl's CCBot page gives the rule User-agent: CCBot / Disallow: /, and its FAQ says the crawler will then stop crawling your site and also honors Crawl-delay.

What neither source says is that blocking removes content already collected. OpenAI's documentation describes the GPTBot rule as a signal about access, and Common Crawl's FAQ talks about stopping future crawling and points to an opt-out registry. Do not assume a robots.txt change reaches backward in time.

Copy-paste rules for each bot are in our guide to blocking AI crawlers in robots.txt. Remember to check your firewall too: a CDN rule that blocks "AI bots" wholesale can lock out OAI-SearchBot along with the training crawlers.

If your content contains personal information

The robots.txt route covers websites. For personal data about you that appears in model outputs, OpenAI's help article says individuals in certain jurisdictions can object or make data subject requests through its Privacy Portal or by emailing dsar@openai.com, and that some rights are not absolute. That is a privacy process, not a site-level opt-out.

Checklist

• Query two or three recent Common Crawl indexes for yourdomain.com/* and note status codes.
• Search your logs for GPTBot and CCBot and verify IPs against gptbot.json and Common Crawl's ccbot.json.
• Decide on training separately from search, using the table above.
• Read your live robots.txt to confirm the rules you wrote are the rules being served.
• Expect no retroactive removal from robots.txt changes.

What to do next

If you only want the quick answer to "is my site in the public crawl archives at all", Satelo's free site check includes Common Crawl presence alongside crawlability and response speed, with the evidence behind each number, in about 30 seconds. To see which AI bots your robots.txt and server let in right now, run the AI crawler checker. And if your real concern is being found rather than being trained on, the difference between the two disciplines is laid out in GEO vs SEO.

FAQ

Can I find out if my website was used to train ChatGPT?

Not directly. OpenAI does not publish a list of training sites or a lookup tool. You can check indirect evidence: GPTBot visits in your server logs and whether your pages appear in Common Crawl, which OpenAI used for GPT-3.

Does ChatGPT use Common Crawl?

OpenAI's GPT-3 paper lists filtered Common Crawl as 60% of that model's training mix. OpenAI's current help article describes categories of data (public web content, partner data, user and trainer data) without naming specific datasets, so current use is not documented.

Does blocking GPTBot remove my content from ChatGPT?

OpenAI says disallowing GPTBot signals that your content should not be used for training. Its documentation does not say that blocking removes content already collected, and blocking GPTBot does not affect ChatGPT search, which uses OAI-SearchBot.

How do I check if my site is in Common Crawl?

Query the free index at index.commoncrawl.org for a recent crawl with url=yourdomain.com/* and output=json. Each result line is a captured page with its status code. The API is rate limited, so run one query at a time.

What is the difference between GPTBot and CCBot?

GPTBot is OpenAI's crawler for content that may be used in OpenAI's model training. CCBot is Common Crawl's crawler, which builds a public archive used by many researchers and AI developers. They are controlled separately in robots.txt.

If I allow OAI-SearchBot, will my content be used for training?

OAI-SearchBot is for search. OpenAI says that if you allow both OAI-SearchBot and GPTBot it may use one crawl for both purposes, so disallow GPTBot if you want search visibility without training.

Read next

Can ChatGPT Read JavaScript Sites? How to Test What OpenAI's Crawlers SeeCan ChatGPT crawl JavaScript? What OpenAI documents, what independent crawler data shows, a curl test you can run today, and fixes for client-side sites.

ChatGPT Ads: How They Work, What They Don't Touch, and Whether a Small Site Should CareHow ChatGPT ads work as of October 2026: who sees them, how they're bought and ranked, whether they affect answers, and what small sites should do instead.

ChatGPT Atlas Is Retired: What It Means for Site Owners and What to Do InsteadChatGPT Atlas is retired. What that means for SEO: what still decides ChatGPT visibility, how ChatGPT's cloud browser reaches your site, and what to check now.

ChatGPT SEO Services: An Honest Buyer's Guide (What to Pay For and What to Do Yourself)A buyer's guide to ChatGPT SEO services: what agencies actually deliver, what no one can promise, red flags, questions to ask and what you can do yourself.