Meta’s new terms let it crawl business sites for AI training
Meta’s revised Commercial Terms, effective October 30, 2026, add a clause that authorises Meta to crawl websites associated with any business using its products and to use that content to train and develop Meta products. Acceptance is passive: keep using a Facebook Page, an ad account or Meta Business Suite after that date and you’ve agreed. Meta says the change is about product data, not scraping news sites.
What happened
- The new paragraph in Section 1 lets Meta, or anyone acting for it, use scrapers, bots and crawlers, and create test credentials, to access and periodically re-crawl domains associated with a business’s products or services, plus URLs made available through business or user activity on Meta’s products. Listed uses include training Meta products, as detailed by PPC Land.
- Nieman Lab first reported the change on October 15 and asked three legal experts to review it. Attorney Matt Topic told the outlet that “publishers should assume that it would be enforceable,” even though court enforcement is uncertain.
- Meta spokesperson Andy Stone called the reports false on X, saying the update is meant to pull accurate product information from third-party sites for ad and content recommendations, and that robots.txt still lets owners control access.
- As Search Engine Journal points out, Meta’s Self-Serve Ad Terms already let it crawl advertisers’ domains for uses including training. What’s new is the reach: the grant now covers businesses that never buy ads.
Why it matters
This is a contract, not a crawler policy. Meta’s public statements describe intent; the clause itself contains no limitation to product pages, and neither the Commercial Terms nor the updated ad terms mention robots.txt. The legal experts Nieman Lab consulted said this gives Meta an opening to argue the terms override other blocking signals. Meta says it won’t use them that way, but nothing in the text guarantees that.
For SEOs, the technical layer matters as much as the legal one. According to Meta’s crawler documentation, as summarised by SEJ, Meta-ExternalAgent crawls both for foundation model training and for product indexing under a single robots.txt token, so you can’t allow one purpose and refuse the other. Meta-WebIndexer feeds Meta AI search answers and citations, which is where any visibility upside sits. Meta-ExternalFetcher and FacebookExternalHit may bypass robots.txt in certain situations (user-requested fetches, security checks).
For publishers, almost all of whom run Meta business accounts, the choice is binary: accept the terms or stop using Meta’s business products. For e-commerce sites, the practical effect may be modest if you already supply catalogue feeds, but expect Meta to read your product pages more actively, especially as in-app product browsing in ads switches on October 28. In Europe, SEJ reports the European Publishers Council is writing to the European Commission about the terms.
What to do
- Audit robots.txt per Meta user agent. Decide separately on Meta-ExternalAgent (training plus indexing) and Meta-WebIndexer (Meta AI search citations). Blocking the first while allowing the second is a common middle path.
- Snapshot your current setup. Save dated copies of robots.txt and any AI-use declarations before October 30, so your stated preferences are documented.
- Check server logs for Meta crawler activity now and after October 30 to see whether volume or behaviour changes.
- Watch registration and paywall systems for unusual test or automated sign-ups, since the clause explicitly allows test credentials.
- Map your exposure. List every Page, ad account, catalogue and Business Suite asset tied to your brands or clients; all of them count as business use.
- Loop in legal if you publish licensable content, and monitor whether Meta narrows the wording before the effective date.