November 19, 2026
by

Block AI Training on Your Website? A Simple Decision Rule

Block AI Training

Block AI training on your website or stay open to it? Now you get to choose.  

Previously, AI models have been trained on the web and you had no choice over whether they used your website to learn or not. Now, you do, after Anthropic, Google, Microsoft and OpenAI have agreed to support a shared opt-out signal.

Here is what the companies agreed and how to decide what's best for your organisation.

What does it mean to block AI training?

Blocking AI training means telling AI companies they may not use your website's content to teach their models. You state the preference with a signal in your site's robots.txt file, the file that already tells search crawlers where they may go.

It does not remove you from search, it also does not stop an AI tool reading your pages to answer a question, which Microsoft calls grounding. You can block one and allow the other.

Four AI companies agreed one opt-out signal

Thurrott's report of 30 October says the four companies will support a common signal that lets website owners state their preference on the use of their content for foundational generative AI training. An IETF working group created the system, the group will propose it as an international standard once complete.

Anthropic has already implemented it. Google, Microsoft and OpenAI will follow in 2027.

Search Engine Roundtable reports that Google will support the signal in early 2027 through a shared 'policy token' in robots.txt. It adds that existing controls such as Google-Extended will keep working.

The work is still a draft at the IETF, one participant says significant disputes remain. Google's own announcement is Supporting a shared AI training control for websites.

The opt-out is separate from search indexing and archiving. Microsoft says sites can choose training, grounding or both while remaining eligible for Bing search results, grounding means an AI tool using your pages to answer a question.

Block AI training or stay open: the trade-off

Blocking training stops future models learning from your content, staying open lets AI tools learn about your business and mention it in answers.

Those aims pull in opposite directions as the sources show that training and grounding are separate choices.

One critic calls the change 'largely symbolic', because the companies collected content at scale years ago. That view deserves weight, the signal is a statement of preference and it does not change how companies have already used content.

Block AI training on your website: a simple decision rule

In our view, three types of business face three different answers.

You sell original content. Research, paid reports, courses and proprietary data have direct value. Opting out of training makes sense for those pages.

You sell services and want to be found. Your site exists so that others quote and recommend you. Staying open to grounding helps an accountancy practice or consultancy appear accurately in AI answers. Our post on AI governance as a competitive edge explains why clear rules help you present your expertise with confidence.

You hold sensitive material. Client portals, internal documents and anything with personal data should never sit on a public page. Our AI compliance support covers the policies behind that.

Most businesses will sit in the second group with a few pages in the first.

Prepare now even though the signal arrives in 2027

A small amount of preparation pays off. The signal sits in robots.txt and the draft may change, so there is little to install yet. You can audit your site today.

List the content that has real value, the content you want quoted and the content that should not be public at all, then decide which group each page belongs in. When the specification is final, your developer can apply the signal without a debate.

Document the decision too, our AI policy training guide explains how a written rule keeps a team consistent. An AI roadmap shows where this decision sits against your other priorities.

This is a Committed-stage decision

Website content rules belong in your AI policy, which makes this a Committed-stage decision on our AI Confidence Journey. At Confused, take the free readiness assessment first. As independent advisers who do not sell software, we advise on what protects and promotes your business whichever vendor is involved.

Frequently asked questions

Can I block AI training on my website today?

Anthropic has implemented the shared signal already. Google plans support in early 2027 and Microsoft and OpenAI follow in 2027. Existing controls such as Google-Extended keep working in the meantime.

Will blocking AI training hurt my search ranking?

The sources say the opt-out is separate from search indexing. Microsoft says sites can opt out while remaining eligible for Bing results.

What is the difference between training and grounding?

Training teaches a model from your content. Grounding means an AI tool uses your pages to answer a question, as Microsoft describes it.

Does the opt-out remove content already used?

The sources do not say so. One critic describes the change as largely symbolic for that reason.

Decide once and write it down

To block AI training on your website or to stay open is a business decision. Sort your pages by value and visibility now so that you are ready for the 2027 switch.

Not sure whether to block AI training? Take our free AI readiness assessment: it takes two to three minutes, carries no obligation and gives you a readiness score with a personalised report.

Share this post

Subscribe to our AI newsletter

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.