Meta Deletes the Word "Train" From Its Business Terms, and Publishers Call It Half a Win

Meta Deletes the Word "Train" From Its Business Terms, and Publishers Call It Half a Win

Meta pulled the clause before it took effect

Meta edited its commercial terms page on October 17, deleting the passage that had alarmed news publishers. The Verge and TechCrunch both reported the change, which followed a review of the original wording by three legal experts at Nieman Lab's request. The revised terms keep their October 30 effective date and replace a version dated March 6, 2024 — so the edit landed before the deadline, not after it.

The sentence that disappeared

The deleted text said Meta may use any content from domains, web pages or URLs "associated with" a business account "to provide, develop, improve, research, train, and maintain the integrity of" Meta's products and services. One word did most of the work: train. It no longer appears, and training is no longer among the listed uses of content collected by Meta's crawlers.

Crawling rights themselves were never withdrawn

What remains is the grant that lets Meta deploy "automated software (such as web scrapers, bots, robots, spiders, crawlers, and user agents)" plus "test credentials" to access, index, cache, copy and periodically re-crawl domains tied to business accounts. What narrowed is the list of purposes: advertising and business products, plus maintaining the integrity and security of Meta products. Search Engine Journal noted the same crawl permission also sits in the revised Self-Serve Ad Terms, effective the same day.

"We've heard you": Meta says nothing changed

"We've heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we've clarified that in the terms themselves," spokesperson Daniel Roberts told The Verge. "It does not change how Meta uses web crawlers." Content owners can still use robots.txt to signal preferences, he said, and "this is not about scraping news sites." Spokesperson Andy Stone had made a blunter version of that argument on X a day earlier, calling the criticism false.

No exit before October 30

Any business that keeps using Meta's products commercially after October 30 is treated as consenting; the only alternative Meta offers is to stop using those products for business. Crypto Briefing read the timing as a sign Meta preferred fixing the wording to defending it through a long public fight.

Publishers: a narrow win, not a reversal

Sam Quigley, spokesperson for the News/Media Alliance, told the Mercury News the new language "just limits what Meta is allowed to use publisher content for, but not their ability to crawl and get the content in the first place." Training was taken out of the listed uses but not expressly banned, he said: "It's still potentially on the table." The terms also never say Meta will respect robots.txt.

Lawyers say the text still points at training

James Grimmelmann, a professor of digital and information law at Cornell, told Nieman Lab the earlier wording amounted to permission "to train AI on any content from their sites and possibly to deploy some of that content, even in competition with the publishers themselves." He flagged "test credentials" as the most worrying phrase, because it could let Meta sign up as a fake user to get past a paywall. Matt Topic, a media and intellectual property attorney at Loevy & Loevy, called the original a "remarkable provision to bury in the terms," covering material that would likely cost millions of dollars in a licensing market. Steven Lieberman, who represents the Mercury News and seven other MediaNews Group publications, called the move overreaching.

What to watch next

Two things. First, whether Meta's crawlers behave any differently after October 30, and whether respect for robots.txt finally makes it into the text itself — the question Quigley keeps raising. Second, the litigation: TechCrunch noted Meta has defeated a high-profile copyright case, while other suits against AI companies, Meta included, continue. Deleting a word gives publishers a document to keep arguing about, not an exemption.