Meta changes its Commercial Terms: What crawling your website means

You have not signed anything, ticked any box or answered any email. Even so, new rules for how your company handles its own website may apply to you from October 30, 2026. The trigger is Meta’s revised Commercial Terms, the terms and conditions for companies that run Facebook or Instagram advertising or use Meta products in commerce. According to a report by the trade publication PPC Land dated October 17, 2026, the new version contains a clause under which Meta may use automated software to visit, index, cache, copy and re-fetch websites that are linked to a company’s products, and may use the content to train its own products. Consent is deemed given as soon as a company continues to use Meta’s products.

For a company in the German-speaking Mittelstand that runs Meta advertising, this is no minor matter. But it is not a reason to panic either. This article sorts out what is documented, what is not, and which internal steps you can still plan before the deadline. It is not legal advice. The legal assessment belongs with your lawyer.

What is known: the change to the Commercial Terms

The main source for this article is the piece “Meta grants itself the right to crawl advertisers’ websites for AI” by Luís Rijo on PPC Land, published on October 17, 2026. PPC Land reports that the change was first described by the journalists Andrew Deck and Hanaa’ Tameez at Nieman Lab on October 15, 2026. We were not able to retrieve the Nieman Lab article itself. Everything we say about its content comes from PPC Land’s account.

The key points, as PPC Land describes them:

• Effective date: The new terms apply from October 30, 2026. They replace a version dated March 6, 2024.
• The clause: A new provision in Section 1 allows Meta to use automated software and test credentials to access websites that are linked to a company’s products. Meta may index, cache, copy and re-fetch them, and use the content to train its own products.
• Consent: It is passive. Anyone who continues to use Meta’s commercial products agrees by doing so.
• Opt-out: The report does not describe a partial opt-out.
• robots.txt: According to the report, the terms do not mention it.
• Crawler: Meta-ExternalAgent is described as a mixed-purpose agent, meaning not a crawler for a single purpose.

Besides this clause, PPC Land lists further changes: a new section on AI features (Section 5), provision “AS IS”, an expanded indemnification obligation, five supplemental terms under Section 8, and a clause 9(g) that allows Meta to discontinue products. Liability is said to be capped at the greater of 100 US dollars or the fees paid in the last twelve months. For disputes, individual arbitration is said to apply to US companies, while outside the US courts in California or Ireland could come into question.

What we were able to check ourselves

We retrieved the page facebook.com/legal/commercial_terms ourselves. It carries a notice: “We’re updating the Meta Commercial Terms. These updates will go into effect on October 30, 2026.” The version currently displayed there is dated March 6, 2024. This supports the effective date and the date of the old version.

We did not find the text of the new clause on this page. The version currently displayed mentions neither crawling of websites nor AI training. Our description of the clause therefore rests on PPC Land’s report and its rendering of the Nieman Lab research, not on a full text that we have read ourselves. We point this out because it makes a difference whether you have read a clause yourself or are relying on a description of it. Read the wording as soon as it can be viewed on Meta’s site, or have your legal counsel read it.

Three voices, three readings

The report reproduces three assessments, each suggesting a different reading. We present them as statements by the people named, not as established facts.

Meta itself: The spokesperson Andy Stone said on October 16 that the clause is narrower than critics believe. It is meant to ensure that Meta can retrieve “accurate information about products for sale”, that is, correct information about products that are for sale, in order to improve ads and recommendations. This is the company’s account. It has not been independently verified.

A critic: Jason Kint, CEO of Digital Content Next, reads the wording as Meta granting itself the right to crawl “websites across the open web”. This is his assessment of the text.

A lawyer: According to the report, the attorney Matt Topic of the firm Loevy & Loevy says it is unclear whether a court would enforce the clause. Publishers should nonetheless assume that it would be enforceable.

When a provider, a critic and a lawyer read the same clause differently, the only safe stance is this: do not guess, check.

Availability: This is not a feature, it is a contract change

A different logic applies here than with a new product. There is nothing to “activate” and nothing you could try out. This is an announced change to terms of use, effective October 30, 2026. Meta’s own page confirms the effective date.

Whether the change also applies to companies in Germany, Austria and Switzerland depends on the contractual relationship. The report names no country-specific restriction and mentions courts in California or Ireland for disputes outside the US. This suggests that companies in the DACH region may also be affected. We cannot say so conclusively, because the report does not name which products exactly count as “commercial products”. That, too, is one of the points you should clarify before the deadline.

Unlike with a beta product, there is no test phase here in which you could observe what happens. The deadline is fixed. Anyone who does nothing until then has decided to accept the new terms by continuing to use the products, whether knowingly or not.

What this means for a company with 10 to 250 employees

The question is not whether Meta reads your website. The question is who in your company makes this decision. In many businesses, the marketing lead or an agency runs the ads, the account is paid for with some company card, and nobody has ever read the terms and conditions. A contract change like this one is then not decided, it simply happens.

Three things are practically relevant for you:

• Which content is tied to Meta? Everything you link to a website in ads, shops or catalogs belongs here: product pages, price lists, landing pages, online shop. For companies with a catalog or shop, the material is often more extensive than one assumes.
• Which of this content is sensitive? Prices for large customers, drafts, pages with restricted access. According to the report, the clause mentions test credentials. What that means in practice is open. You should know whether there is content behind a login that you do not want to see in someone else’s training data.
• Are there alternatives? According to the report’s account, anyone who does not want to accept the terms can only stop using Meta’s products. Whether that is realistic for your company depends on how much of your revenue depends on Instagram and Facebook.

An everyday example

A manufacturer of workwear with 60 employees runs ads on Instagram and operates a shop whose catalog is connected to the Meta account. Alongside the public catalog, there is a protected area for dealers with individual tiered prices. The marketing manager set up the account years ago, and the agency maintains the campaigns. Shortly before the deadline, the managing director asks: “Who actually decided that Meta may read our pages?” Answer: nobody. It just happened.

This is a thought experiment, not a statement about a customer and not a figure from the report. But it shows how such a change arrives in practice: not as a decision, but as a gap between marketing, management and legal.

Mapping: Nadja, Natascha and the approval step

On amazing-agents.com, we describe Nadja in the Research team, who handles research, topic research, competitor scans and visibility in AI answers. In the Social Media team, Natascha works on the principle that nothing goes live without your approval, and Magdalena checks every draft as QA and brand check before it goes to approval. For the case of the Meta terms, the roles can be set side by side like this:

• Nadja (Research) corresponds to the risk assessment: What does the report say, what does the source say, what is open? Which pages of your website are visible to AI systems? The result is a briefing, not legal advice.
• Natascha (Social Media) corresponds to the operational review of the channels: Which content, campaigns and shop links are tied to your Meta account? Deciding and approving stays with you.
• Magdalena (Brand Check) corresponds to the review step between draft and publication, in this case: before you decide internally whether and how you accept the terms.

This is a side-by-side comparison of tasks, not a product promise. Meta’s terms are someone else’s contract. Our agents prepare reviews; they do not take the decision off your hands and they do not give legal advice. The question you have to answer either way is: Where is the approval step?

Agent with limits and an approval step: without and with

Without an approval step: The contract change takes effect because nobody reviewed it. Consent is given by continued use, according to the reports without a click and without a follow-up question. Whether your management knows about it, whether your legal counsel has read it and whether your data protection officer was involved only becomes apparent in a dispute. This is exactly what the report describes as the mechanism: platforms acquire rights through terms rather than through product announcements.

With an approval step: Before the deadline, a short briefing is ready: What changes, which content is affected, which options exist? A named person decides whether you continue to use the Meta products, with which restrictions and with which legal review. The decision is documented. This is the same idea as with our agents: an agent prepares, a human approves. On our website, we describe it like this: every agent runs within a defined framework with approval steps for critical actions; emails, for example, are created as drafts and not sent directly.

With Meta’s new terms, according to the available reports, there is no approval step on the company side. Such a step would have to be created internally. That is exactly the task for the coming days.

How we proceed in an audit

We are not claiming results from customer projects here, but describing how we proceed when a company wants to examine its dependence on an advertising platform:

1. Capture accounts and links. Which Meta accounts, shops and catalogs exist, who has access, which websites are linked?
2. Sort content by sensitivity. Public, internal, protected. We separate what the world can read anyway from what should not end up in outside hands.
3. Read the sources. We read the terms and the reliable reports about them, note where statements come from providers, and mark gaps.
4. Present options. Continue, restrict, examine alternatives. With the consequences for revenue, effort and risk, as far as they can be substantiated.
5. Document the decision. Who decided what, when, and on what basis? This is your evidence in case someone asks later.

The result is a decision proposal, not a promise. Legal advice stays with your lawyer.

What you should prepare before October 30

You have just under two weeks from today. These steps can be completed in that time:

1. Name a responsible person. Clarify who in your company is responsible for the Meta accounts, and record that person and a deputy. Without a name, there is no decision.
2. List linked content. Collect which websites, shops and catalogs are connected to your Meta account. Mark content behind a login or with confidential prices.
3. Read the wording, or have it read. As soon as the new Commercial Terms can be viewed on Meta’s site, have your legal counsel review the clause in Section 1. Ask specifically: What does “commercial products” mean for us, and what effect does consent through continued use have?
4. Technical questions for IT. The report notes that the terms do not mention robots.txt. Ask your IT team or agency which crawlers are currently allowed to see your site, and have them explain what a block achieves and what it does not. Do not rely on a technical block settling the contractual question.
5. Involve data protection. Discuss with your data protection officer whether personal data sits on linked pages and how the change fits with that.
6. Make and document the decision before the deadline. Continue, restrict, or examine whether alternatives come into question. Record who decided.

What the source does not say

The report does not state which Meta products count as “commercial products” and which products are supposed to work with the training data. It does not explain what “test credentials” means in practice and how Meta would access pages behind a login. It gives no information on retention period, deletion or handling of the training data, no review by an authority or a court, and no statement on whether the text of the terms matches Meta’s actual practice. There is also no indication of how a company could object other than by terminating. And: the wording of the clause itself was not available to us. We have therefore limited every statement in this article to what is documented, and have named the gaps.

Conclusion

Meta’s new Commercial Terms show how platforms acquire new AI rights: not through a product announcement, but through a change to terms and conditions that applies through continued use. Whether the clause reaches as far as critics say, or is narrower, as Meta claims, this article cannot answer for you. It can, however, show you what you can do before October 30: clarify responsibility, capture linked content, have the wording reviewed and make the decision consciously. The difference lies not in whether you use Meta, but in whether someone has decided: an agent with limits and an approval step instead of a consent that simply happens.

Sources

• PPC Land, Luís Rijo: “Meta grants itself the right to crawl advertisers’ websites for AI”, October 17, 2026, ppc.land (retrieved on October 17, 2026)
• Meta: Commercial Terms with notice of the change effective October 30, 2026, facebook.com/legal/commercial_terms (retrieved on October 17, 2026; the version displayed is the one dated March 6, 2024)
• Nieman Lab, Andrew Deck and Hanaa’ Tameez, October 15, 2026 (considered only via PPC Land’s account, article not retrieved)
• Amazing Agents: descriptions of Nadja, Natascha and Magdalena and of the approval framework, amazing-agents.com