Meta has struck the word “train” from the clause that lets it crawl business customers’ websites, but the permission to crawl survived the edit. The revised Commercial Terms still take effect Oct. 30. Search Engine Journal’s Matt G. Southern reported the change on Oct. 19, building on Nieman Lab’s Oct. 17 coverage.
Our earlier coverage described a clause reaching any business that uses Meta products, advertiser or not. That reach has not shrunk. A company that only manages a Facebook Page remains covered.
The original wording allowed Meta to use collected content for a list of purposes that ended with training its products. The revision replaces that list with a narrower frame: advertising and business products, including tools that help a business build ads and connect with interested users. An integrity purpose stays, now phrased around the security of Meta’s products as well.
The access mechanics were left alone. Meta may still run automated software, may still “create test credentials” to gather information, and may still return periodically to pick up site updates. The new language makes no mention of robots.txt.
Regional documents also disagree. The UK English preview of Meta’s ad purchasing agreement, titled Self-Serve Ad Terms, was still carrying the earlier purpose list, training included, as of Oct. 19. The U.S. English version did not.
Meta’s explanation came from spokesperson Daniel Roberts, who told The Verge the consolidated crawling terms only ever covered ads and business products, and that the edit spells this out. He said the update leaves Meta’s use of web crawlers as it was and that “This is not about scraping news sites.” To Nieman Lab, the company said the first draft “wasn’t clear” about honoring robots.txt preferences, and that the updates grant no “new rights that supersede those preferences.”
The lawyers who reviewed the original wording are not persuaded. Attorney Matt Topic of Loevy & Loevy, who works on media and IP matters, said “develop” is broad enough to include “train,” and noted that “business product” does not look like a defined term. James Grimmelmann, a professor at Cornell, said the edit leaves the “associated with your products or services” phrase untouched, which he reads as covering entire sites. He described the purposes wording as “still broad enough to drive a truck through.”
A third reviewer, Benjamin Sobel, a law professor at the University of Wisconsin, addressed Meta’s robots.txt promise directly. The file expresses a site’s preferences, he said. Compliance is voluntary, and the file confers no enforceable legal right.
That is the gap Meta’s assurances do not close. Meta points businesses to robots.txt as the control for which areas get crawled, and its crawler documentation lists separate crawlers for AI training and for Meta AI search, so a site can block one and admit the other. The revised text does not say whether such rules shrink the permission a business gives on acceptance. A promise made to reporters and a clause signed by customers are different instruments.
Pressure from publishers predates the edit. Ahead of the revision, the European Publishers Council voiced opposition to the first wording, and the UK’s News Media Association issued a statement on Oct. 16 condemning it. Southern’s report does not say whether either group considers the revision sufficient.
For site owners, deleting one word narrows a stated purpose, not the technical reach. The crawler can still visit, and the contract, not robots.txt, is the document that would matter in a dispute. Meta has also changed the preview since first posting it, so a clause you reviewed last week may no longer match.
Before Oct. 30, re-read the final clause in both regional versions, check how your robots.txt treats each Meta crawler, and have counsel decide whether “develop” and “business product” are acceptable terms to sign.
Originally reported by Search Engine Journal (Matt G. Southern, October 19, 2026), citing Nieman Lab’s October 17, 2026 report and Meta spokesperson comments to The Verge.
