Here is the short answer: Notion Sites gives you no robots.txt file to edit, so you cannot block individual AI crawlers like GPTBot or ClaudeBot the way you can on a self-hosted site. What you do get is a "Search engine indexing" toggle in the Publish menu. Turn it off, and Notion tells search engines not to index the page, which keeps it out of Google results and sharply reduces the chance an AI crawler finds and uses it. Turn it on, and the page is discoverable on the web, which means Google can index it and AI search crawlers can fetch it. If a page must never reach an AI system, do not publish it as a Notion Site at all: keep it invite-only, or move it to a site you control. This guide walks through the toggle, the limits behind it, and the options that actually protect sensitive content.
How Notion Sites Work
Notion Sites turns an ordinary Notion page into a live website. You open the page, select Share, open the Publish tab, and select Publish. The page gets a public address on a notion.site domain, and anyone with the link can read it. As of October 2026, all plans can publish pages and claim one notion.site domain, and paid plans add custom domains, SEO settings, themes, and Google Analytics.
Publishing a site is a separate thing from the "Anyone on the web with link" setting in the Share menu. A page can be shared with a link without being published as a site, and a published site can exist without search engines knowing about it. That distinction is the whole ballgame for AI crawler control, because the two settings do different jobs.
The first setting, link sharing, decides who can open the page. The second setting, the Publish tab's "Search engine indexing" toggle, decides whether search engines are invited to find it. AI crawlers are not search engines, but they discover pages the same way search engines do: by following links and by reading search indexes. A page nobody can find is a page no crawler will fetch. A page in Google's index is a page every AI search crawler can reach.
The Search Engine Indexing Toggle: What It Actually Does
In the Publish menu, the "Search engine indexing" option controls whether your site is discoverable on the web. When it is on, Notion allows search engines to crawl and index the page. When it is off, Notion tells search engines not to index it. That is the closest thing Notion Sites has to a robots.txt rule, and it is the first setting to check.
Here is the honest part: the toggle is a request, not a wall. It works the same way a noindex tag works. Well-behaved crawlers, including Googlebot and the major AI crawlers, respect it. But a scraper that ignores indexing signals can still fetch a public page if it has the link, because the page is still publicly readable. The toggle removes the page from the discovery pipeline. It does not put a lock on the door.
There is also a timing caveat. Notion says sites can take up to four weeks to be indexed and appear in search results once discoverability is on. The reverse is also slow: turning indexing off does not instantly remove a page from Google's index or from an AI search index. Search engines remove pages on their own schedule, which can take days or weeks. If your page was already indexed and cited, flipping the toggle is the start of a removal, not an instant erasure.
Step by Step: Turning Indexing Off
The steps are short, but they are easy to get wrong, because the Publish tab and the Share menu look similar and do different things. Follow them in order:
- Open the page you want to protect in Notion.
- Select Share at the top of the page.
- Open the Publish tab. This is not the same as the "Share to web" link setting.
- Find Search engine indexing and make sure Discoverable on the web is turned off.
- Check the general access setting too. If the page still says "Anyone on the web with link," decide whether that is what you want. If the content is sensitive, remove public access entirely and invite specific people instead.
- Repeat the check for every published page. The toggle is per page, not per workspace. A setting you changed on your homepage does nothing for the ten other pages you published last year.
On paid plans you can also edit the link title and description that search engines show, which is useful for the pages you do want indexed. That is an SEO feature, not a privacy feature, but it is worth knowing about so you do not confuse the two menus.
What Notion Does Not Let You Do
There is no robots.txt editor in Notion Sites. You cannot add a rule that names GPTBot, ClaudeBot, or PerplexityBot. You cannot add a rule that blocks one crawler while allowing another. You cannot upload a llms.txt file to a notion.site domain, and you cannot set custom response headers. The platform gives you one binary switch, discoverable or not, and that is the entire control surface.
Custom domains do not change this. A paid plan lets you point your own domain at a Notion Site, which is good for branding and for SEO, but the page is still served by Notion, and the indexing toggle is still the only crawler control you have. If a guide tells you to edit the robots.txt of your Notion Site, it is describing a different platform. There is no file to edit.
Enterprise workspaces get one additional control: workspace owners can disable publishing sites, forms, and public links entirely, from Settings and then Security. Turning that on takes down sites that are already live. That is a workspace-level kill switch, useful for a company that decides no Notion content should be public at all, but it is not a per-page AI opt-out, and it is not available on personal or small-team plans.
The Stronger Options, in Order
1. Unpublish the Page
If a page should not be on the public web at all, unpublish it. Open the page, select Share, open the Publish tab, and select Unpublish. Then check the general access setting and remove "Anyone on the web with link" if it is still there. An unpublished page with link sharing removed is visible only to people you invite. That is the strongest protection Notion offers, because it removes the page from the web rather than asking crawlers to ignore it. Use this for anything with client data, unpublished research, internal docs, or personal information.
2. Keep Sensitive Pages Invite-Only From the Start
The cleanest habit is to never publish sensitive content in the first place. Notion's sharing model makes this easy: invite the specific people who need a page, and leave public sharing off. A page that was never published and never link-shared has no public URL for a crawler to find. The most common mistake is the reverse: someone publishes a page to share it with one colleague, forgets it is public, and it sits discoverable for a year. Audit your published pages twice a year. The Publish tab lists what is live, and the list is usually longer than people expect.
3. Set an Expiry on Shared Links
Notion added auto-expiring public pages in October 2022. On a paid plan, you can set public sharing to stop after an hour, a day, a week, or on a custom date. This is genuinely useful for templates, proposals, and time-limited shares: the page stops being public on its own, even if you forget about it. It does not remove content already copied or indexed, but it closes the window for future crawling. If you share Notion pages with clients or customers, set an expiry every time. It costs nothing and it removes the most common failure mode, which is forgetting.
4. Move the Content to a Site You Control
If AI training is a red line for a piece of content, Notion Sites is the wrong home for it. Publish it on a domain you own, where you can write a robots.txt file that names the training crawlers and tells them to stay out, while keeping the AI search crawlers that send you traffic. The standard file looks like this:
User-agent: GPTBot Disallow: / User-agent: ClaudeBot Disallow: / User-agent: CCBot Disallow: / User-agent: Google-Extended Disallow: /
That split, training out and search in, is the decision most site owners actually want, and it is the one Notion cannot express. Our free robots.txt generator builds this file from checkboxes, and our robots.txt tester verifies it before you publish. If you are weighing whether to block at all, read our guide on the pros and cons of blocking AI crawlers first, and the complete list of AI crawlers to see what each bot does.
What About llms.txt on Notion?
You cannot add a llms.txt file to a Notion Site, because you cannot upload files to the root of a notion.site domain. That is worth knowing, because llms.txt is often presented as the friendly alternative to blocking: a file that tells AI systems what your site is about. On a site you control, it is a cheap, useful addition. On Notion, it is not available at all. If you want AI systems to understand your content rather than just crawl it, that is another reason to keep your public-facing content on your own domain, and to use Notion for what it is best at: private collaboration and drafting.
The Indexing Decision, Made Simple
Most Notion users only need to answer one question per page: is this page meant to be found by strangers? If yes, turn search engine indexing on, write a good title and description, and let Google and AI search do their work. If no, turn indexing off, or better, do not publish the page at all. There is no middle setting, and there is no per-crawler setting, so the decision is binary. The pages that cause trouble are almost always the ones in the middle: published for one person, indexed for everyone, and forgotten.
Do a quick audit today. Open Notion, go through your published pages, and for each one ask whether a stranger finding it in Google would be fine. For the ones where the answer is no, turn indexing off or unpublish. It takes ten minutes, and it is the highest-value privacy step available on the platform.
Frequently Asked Questions
Can I block GPTBot specifically on a Notion Site?
No. Notion Sites has no robots.txt editor, so you cannot name individual crawlers. The only control is the search engine indexing toggle, which applies to all search engines at once. To block GPTBot while allowing other crawlers, you need a site you control.
Does turning off search engine indexing stop AI crawlers?
It sharply reduces the chance they find and use the page, because it removes the page from search indexes, which is how most crawlers discover content. It does not put a technical block on the page. A crawler that already has the link can still fetch a public page, so for content that must never reach an AI system, unpublish it or keep it invite-only.
How do I unpublish a Notion Site?
Open the page, select Share, open the Publish tab, and select Unpublish. Then check the general access setting: if it still says "Anyone on the web with link," open that dropdown and select Remove. Both steps matter. Unpublishing alone leaves the link-shared page readable.
Can I use a custom domain to control robots.txt on Notion?
No. A custom domain on a paid plan changes the address of your Notion Site, but Notion still serves the pages and the indexing toggle remains your only crawler control. There is no robots.txt file for you to edit on any Notion plan.
Does the indexing toggle affect pages I already published?
Yes, but not instantly. Turning indexing off tells search engines to stop listing the page, and they remove it on their own schedule, which can take days or weeks. Notion notes that indexing itself can take up to four weeks, so removal is not immediate either. If a page was already cited by an AI answer, the toggle does not erase the citation.
What is the safest way to share a Notion page with one person?
Invite that person from the Share menu and leave public sharing off. Do not publish the page and do not enable "Anyone on the web with link." On a paid plan, also set the link to expire. That combination, invite-only plus an expiry, is the strongest sharing setup Notion offers short of unpublishing. Our Substack guide covers the same decision on another hosted platform, and our Ghost guide shows what the same decision looks like on a platform that does give you a robots.txt file.
Notion Sites is a publishing tool with one privacy switch. Learn what that switch does, use it deliberately, and keep anything you truly need to protect off the public web entirely.
