How to Opt Out of AI Training: A Freelancer's Guide to the New Shared Robots.txt Signal

If you make a living from what you publish online — portfolio pieces, case studies, blog posts, photographs, code samples — then a quiet announcement on 30 October 2026 matters to you. OpenAI, Google, Microsoft and Anthropic said they will honour a single shared signal that lets a website owner opt out of AI model training. Instead of configuring a different rule for every company's crawler, you can now write one entry in your robots.txt file and have all four companies read it the same way.

For freelancers, this is genuinely useful — but only if you understand exactly what it does and what it does not do. This guide explains the signal, how to add it, and whether opting out is right for you or your clients.

What Just Changed on 30 October 2026

Until now, opting out of AI training meant keeping a growing list of company-specific crawler rules up to date: one token for Google, another for OpenAI, another for Microsoft, and so on. Miss one, and that company's training pipeline could still include your content. The new approach collapses all of that into a single policy token: __usage_ai-training.

The token comes from an individual Internet-Draft titled "Machine-Readable Training Preference Signal", written by Anthropic standards lead Kevin Kelley and dated 30 October 2026. As an individual draft it is not yet a formal standard and carries no enforcement power: the four companies have committed to honouring it voluntarily — a meaningful step, but a commitment, not a law.

The two lines that do the job

A website owner who wants to opt out adds this block to the robots.txt file at the root of their domain:

User-Agent: __usage_ai-training
Disallow: /

That is the whole mechanism. You can also carve out exceptions with Allow lines — for example, opting out of everything except a public press kit folder. If you have never edited robots.txt before, it is a plain text file that lives at yoursite.com/robots.txt, and most hosting control panels and site builders let you edit it directly.

Who is backing it, and when each company actually starts honouring it

The four companies have not moved in lockstep, and the timelines matter for your planning:

• Anthropic has already begun implementing support for the token.
• OpenAI says it will implement the signal in the coming months, without a fixed date.
• Microsoft targets 1 February 2027, and is adding a separate, more detailed control for AI grounding — the step where live web content is pulled into chat answers, which is different from training model weights.
• Google says it aims to support the token in 2027, and has confirmed that if you use the shared token without separately naming Google-Extended, Google will treat it as blocking Google-Extended too.

The takeaway: adding the token today is about being covered as support rolls out, not an instant switch. Keep existing company-specific rules until each company confirms support.

What the Opt-Out Does — and What It Does Not Do

It does	It does not do
Tell all four participating companies not to use your site's content for training AI models	Stop crawlers from visiting, fetching or caching your pages
Replace four separate company-specific training rules with one shared entry	Remove content already collected into datasets or already trained into models
Let you exempt specific folders with Allow lines	Keep your pages out of search results or out of AI-written answers that quote live pages
Work alongside your existing search and indexing rules	Bind companies outside the four participants, or carry legal enforcement on its own

The most common misunderstanding is the third row. Opting out of training is not the same as opting out of being quoted. An AI answer that retrieves your live page and cites it is a different process from training a model on your content, and this token does not cover it. Microsoft's separate grounding control is aimed at that gap, but it is a different setting with its own timeline.

Should You Opt Out? A Freelancer's Decision Guide

There is no single right answer, and the honest answer depends on how you earn. Work through these three cases before you touch your robots.txt file.

Opt out if…

Your published work is the product itself. If you are a writer, photographer, illustrator, designer or researcher whose portfolio pieces could be absorbed into a model that then competes with you, opting out protects the asset you sell. The same applies to original research, proprietary frameworks and signature case studies that clients pay you to create.

Stay visible if…

Your business depends on being discovered through AI answers and search. If clients find you because an AI assistant recommends your profile or quotes your blog, a blanket opt-out could reduce that discovery over time. Many freelancers want to be found by AI search while keeping core creative work out of training sets — which leads to the middle path.

The middle path: partial opt-outs by folder

You do not have to choose all or nothing. A sensible middle path is to opt out the folders that hold your original creative work — your portfolio, your image library, your premium articles — while leaving your marketing pages, service pages and public writing open. This is exactly what Allow lines are for, and it is the approach most freelancers will find most practical. Document whichever choice you make, because a client may one day ask why their site is set the way it is.

How to Add the Opt-Out to Your Site, Step by Step

1. Find your robots.txt file. Visit yoursite.com/robots.txt. If it exists, you will see its current contents. If it returns a "not found" page, you will be creating it from scratch.
2. Back up the current file. Copy its contents somewhere safe before editing. A mistake in robots.txt can accidentally block search engines, so a backup is essential.
3. Add the shared token block. Add these two lines, keeping any existing rules intact: User-Agent: __usage_ai-training, followed by Disallow: /. Place it as its own block, separate from your other User-Agent groups.
4. Add exceptions if you want a partial opt-out. Under the Disallow line, add Allow lines for folders you want to keep open, such as Allow: /press-kit/.
5. Keep your existing company-specific rules. Because support rolls out through 2027, do not delete your Google-Extended, GPTBot or other existing training rules yet. The shared token is an addition, not yet a replacement.
6. Test the file. Revisit yoursite.com/robots.txt to confirm it loads, and check that your existing search rules are unchanged. If your platform offers a robots.txt validator, use it.
7. Note the date. Record when you added the token and which folders you exempted. When each company confirms support, you can revisit and simplify your file.

If you build client sites, add one step first: ask the client. The opt-out is a business decision about their content, not a technical default to set silently. A short note explaining the trade-off, and their written choice, protects both of you.

What Still Works Today

Because the shared token is still rolling out, the controls that work today remain important:

• Company-specific crawler rules such as Google-Extended remain the rules that are actually enforced right now. Keep them until each company confirms shared-token support.
• Microsoft's existing tags still apply, but Microsoft says its older NoArchive and NoCache tags will stop being read after 1 March 2027, so sites using them should set the new preferences before then.
• Your platform's own settings may offer additional AI controls. Check your site builder or hosting panel for AI training or crawler settings beyond robots.txt.

For freelancers who build and host client websites, having direct access to the site's files makes all of this straightforward. Tools such as Lovable Pro for building AI-powered client sites, Framer Pro for portfolio and client websites, and the Hostinger 3-in-1 Bundle for hosting with direct file access give you the control you need to edit robots.txt and manage these settings for yourself and your clients.

What to Tell Clients About It

Clients will hear about this story in the news and ask you about it. A clear, honest summary works best: four major AI companies have agreed on one shared way to say no to AI training, the mechanism is a simple robots.txt entry, support is rolling out through 2027 rather than live everywhere today, and it does not remove content from existing models or keep pages out of search results. Offer them the three-way choice — full opt-out, stay visible, or partial opt-out by folder — and record their decision. That conversation is itself a service worth offering: many small businesses have no idea robots.txt exists, let alone that it now carries an AI training choice.

The wider debate is far from settled — Meta delayed its new AI-training terms to 20 November 2026 after publisher pushback the same week — so revisit your settings as the draft evolves before its May 2027 expiry.

FAQ

Does the new opt-out stop AI from using my content immediately?

Not immediately everywhere. Anthropic has begun implementing support, but OpenAI, Microsoft and Google roll out through 2027. Keep your existing company-specific rules until each company confirms support.

Will opting out hurt my Google search rankings?

The shared token is a training preference, not a crawling or indexing rule. Microsoft has said explicitly that opting out of training does not affect eligibility for regular Bing results, and the token is designed to separate training permission from search visibility.

Does it remove my work from AI models that already trained on it?

No. Updating robots.txt does not remove content already collected into datasets or already trained into model weights. The token governs future training collection by participating companies.

Can I opt out only part of my site?

Yes. Use Allow lines under the Disallow rule to exempt specific folders, such as a press kit or public marketing pages, while opting out the rest.

Should I delete my old crawler rules now?

No — not yet. The old rules are what is actually enforced today. Add the shared token alongside them, and simplify only as each company confirms support.

Conclusion — the Next Step

The shared AI-training opt-out is a small file change with a large meaning: for the first time, the biggest AI companies have agreed to read the same "no" from website owners. Make a deliberate choice — full, partial or none — for your own portfolio, and offer the same clear choice to your clients. Open your robots.txt file this week, back it up, add the shared token alongside your existing rules, and revisit it as each company's support goes live through 2027. And if you want your freelance profile itself to be easy to verify and trust, a verified listing on DIGGZA puts your credentials in front of clients who check before they hire.