On October 30, 2026 — nine days from now — Meta’s updated Commercial Terms take effect. If your company uses any Meta business products, which is more common than it sounds (a Facebook Page, Ads Manager, Business Suite, a product catalog, measurement services — not just ad spend), the updated terms include a clause authorizing Meta to crawl the websites associated with your business account. The version Meta originally published went further, explicitly permitting use of the collected content to train its products. After a day of public pushback from the publisher industry, Meta removed the training sentence on October 17 — and kept everything else. This post is the decision-relevant summary: what happened, what the revised clause actually permits as reported, which Meta bots are involved, and how to review or opt out before the 30th.

A scoping note on sourcing, because honesty requires it: the terms themselves live on Meta’s legal pages, and the account of the clause and its revision below is as reported by the outlets we verified — AI Affairs (October 16, 2026, summarizing Nieman Lab’s October 15 report: https://www.aiaffairs.com/corporate/meta-commercial-terms-ai-training-business-websites-october-2026/), Unite.AI (October 17, 2026, which verified the change against archived captures of Meta’s preview page: https://www.unite.ai/meta-rewrites-web-crawling-clause-in-pending-commercial-terms/), and TechWyse (October 21, 2026: https://www.techwyse.com/news/platform-updates/meta-commercial-terms-website-crawling-ai-training). Read the final text yourself before the 30th, and get legal advice on your own situation — nothing here is legal advice.

What happened, in four days

  • October 15: Nieman Lab reported that the pending Commercial Terms (effective October 30) would permit Meta and parties it appoints to deploy automated software on domains, webpages and URLs linked to a business user’s products or services — and, as reported, explicitly listed training among the permitted uses of the collected content.
  • October 16: The pushback. The News/Media Alliance called the terms “outrageous and probably illegal” (president and CEO Danielle Coffey, via Unite.AI), arguing any business with intellectual property online and a Meta business account could have its copyrighted material taken. Meta’s Andy Stone responded publicly that there was no change to how Meta uses web crawlers and that, as always, content owners can use robots.txt to manage access.
  • October 17: The revision. An archived capture of Meta’s terms preview page from October 16 preserved the original sentence — “Meta may use any content from such domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of, the Meta Company Products…” — and a capture from October 17 shows it gone, per Unite.AI. The verb train no longer appears in the provision.
  • October 21: TechWyse’s analysis of the end state: “Meta Narrowed the Purpose, Not the Crawling Permission.”

What the revised clause permits (as reported)

Per the reporting above, the revised clause still authorizes Meta — or any entity acting on its behalf — to use automated software (“web scrapers, bots, robots, spiders, crawlers, and user agents”) and to create test credentials to access, index, cache, analyze, collect, retrieve or crawl the domains, webpages and URLs associated with your products or services, or made available through use of Meta’s products. The authorization extends to periodically pulling updates from those domains. The stated purposes are now narrower: to provide, develop, improve, research and maintain advertising and business products, and to maintain the integrity and security of Meta Company Products.

Four properties of that language matter for a B2B decision:

  1. “Train” removed is not “never train.” TechWyse is explicit that the remaining verbs — “develop,” “improve,” “research” — “cannot be treated as proof that Meta intends to use the material as training data,” but equally the clause “does not say that crawled content will never be used in AI-supported business products. The contract leaves that boundary unresolved.” That is the honest state of the document as reported: a real narrowing, and an open question.
  2. The audience is wider than your advertiser team. The terms apply to any use of Meta products for business or commercial purposes — managing a Page, operating a catalog, using Business Suite, accessing measurement services. A company spending zero on ads falls inside, as reported by TechWyse.
  3. There is no partial opt-out. Continued access or use after notice of the update constitutes agreement (Unite.AI). The terms, as reported, do not provide a way to reject only the crawling paragraph while keeping the products.
  4. robots.txt and the contract are different layers. Meta says content owners can manage crawler access through robots.txt (per its spokesperson, via Nieman Lab), and that its revised language does not supersede those signals (per TechWyse). But the Commercial Terms themselves do not mention robots.txt, and Meta’s own crawler documentation — as reported by TechWyse — says some user-requested fetching and certain security or integrity checks may bypass robots.txt restrictions. The clause is a permission between Meta and you; robots.txt is an instruction you give the bots. Both exist, and they do not fully overlap.

The Meta bots, and the actual opt-out mechanism

The terms debate tends to collapse three questions into one — may Meta access a page, which crawler accesses it, and what happens to the content — and the only one you can act on this week is the second. Per TechWyse (citing Meta’s crawler documentation), the relevant agents each have their own identifier:

  • Meta-ExternalAgent — associated with foundation-model training and direct content indexing.
  • Meta-ExternalAds — advertising and other business-related products.
  • Meta-WebIndexer — Meta AI search.
  • Separate agents handle shared-link previews and user-requested page retrieval.

A robots.txt rule addressing one crawler does not control the others. The practical review: identify which of your URLs are tied to Meta (business-Page linked domains, ad landing pages, catalog URLs), then decide per agent — and note the trade-off TechWyse flags: a blanket block “may restrict more than AI training crawlers. It could also affect Meta’s ability to read advertising destinations, build link previews or surface a website in AI-assisted discovery.” A typical B2B SaaS posture — and this is judgment, not a legal position — is to keep the search and ads agents allowed for your public marketing pages and to block or restrict the training agent from anything proprietary: technical docs, customer case-study detail, pricing internals, changelogs that describe unreleased work. If you run your site on Cloudflare, the September 15 crawler-class rebuild we covered earlier this week now enforces that split at the edge rather than hoping the bots comply.

The honest framing: what is actually at stake

This story is being covered as an AI-training story, because the removed word was “train.” For a B2B SaaS site, the realistic frame is different, and smaller in one way and bigger in another.

Smaller: Meta is not currently a significant citation surface for B2B buying decisions. Our cross-surface work — which AI surfaces actually cite B2B SaaS brands — tracks ChatGPT, Perplexity, Gemini and Claude, because that is where B2B buyers evaluate; Meta AI is not among the surfaces we treat as revenue-relevant today, and nothing in this terms change changes that. You are not going to lose “Meta rankings” over this clause, because there is no meaningful Meta answer-citation game to lose yet. (Observational judgment from our runs, not a measured market statistic.)

Bigger: the clause is about what happens to your data, not your visibility — and it reaches every domain tied to a business account, not just the ones you think of as “the website.” If your marketing site, your docs, and a staging or legacy property all sit behind business accounts, all of them are in scope as reported. The decision is a data-protection decision with a visibility footnote: set your crawler preferences deliberately so that what Meta can reach is exactly what you are comfortable with Meta reaching, while keeping the search and preview paths that your legitimate use of Meta depends on.

That is a review, not a panic: one afternoon to inventory your Meta-linked domains, one afternoon for the robots.txt and edge rules, and a re-check when the final text of the October 30 terms is live. The full inventory of AI crawler user-agents and their purposes is in our AI crawlers reference.

Next step: if you want this reviewed as part of an ongoing posture — crawler policy, citation tracking on your money queries, and a documented decision log rather than one-off reacts to each platform’s terms update — that is the scope of what we run for B2B SaaS clients, and the about page explains how the practice works.