Meta’s crawling terms lost the word “train” late on Friday, October 16, but the revised version still takes effect October 30, and publishers say the risk is not resolved. Any business that keeps using Meta’s advertising and business tools after that date accepts the new language, so the decision is live for newsrooms and small companies alike.

The original clause drew fire because legal experts read it as allowing crawls of websites tied to business accounts on Facebook, Instagram and WhatsApp, with the collected content linked to training Meta’s products. Experts interviewed by Nieman Lab argued the reach could extend to nearly every newsroom, since almost all of them hold at least one Meta business account, as AI Weekly summarized. USA 24 has previously reported on the October 30 terms and the possibility that Meta could scrape and train on news sites. This report draws on press coverage of the revised text rather than a direct reading of Meta’s final terms page.

What the revised Meta crawling terms say

The new wording says automated crawling is meant to develop, improve, research and maintain Meta’s advertising and business products, according to Anews. The listed purposes no longer name training, and that is the change Meta is pointing to. The terms still permit scrapers, bots and crawlers to collect content from websites connected to businesses on Meta’s platforms, Rolling Out reported.

The acceptance mechanism matters as much as the wording. Businesses accept the new terms automatically by continuing to use Meta’s commercial advertising and business products, as The Next Web reported. A page owner who keeps running an ad account does not need to sign anything to be bound.

The sequence of events moved quickly:

  • October 16: Meta communications chief Andy Stone posts that nothing has changed in how Meta’s crawlers operate and that site owners can still set preferences through robots.txt.
  • Late Friday, October 16: Meta revises the commercial terms and drops “train” after complaints from business customers and media.
  • October 17: Some outlets report the change, according to CryptoBriefing.
  • October 30: The revised terms take effect.

Why the word “train” mattered to publishers

Meta has signed paid AI licensing agreements with media companies, including CNN and Fox News, according to CryptoBriefing. That history explains the publisher reaction. If Meta can gather material for model development under a general crawl license, it has a cheaper alternative to paying for content. Legal experts cited potential licensing fees worth “many millions” of dollars to publishers, and Slashdot framed the original language as a way to avoid those fees.

The narrowing has a limit that matters. The revised text uses the words develop, improve and research, and those words can cover much of what training involves. Whether the edit changes anything in practice depends on how Meta reads them, and the terms do not spell that out. A verb list can be as broad as the company wants.

What Meta says has not changed

Meta has said its crawling terms only ever applied to ads and business products, and that it has not changed how its crawlers work, as Zetik, citing The Verge, reported. Stone’s October 16 post made the same core point: robots.txt preferences remain the site owner’s main lever.

That framing places the terms in the business-tools category. Publishers counter that almost every newsroom holds a business account, so the clause reaches far beyond the advertising tools Meta describes.

The company also carries legal history into this fight. Meta beat a high-profile copyright lawsuit last year, but it still faces broader legal scrutiny over how it gathers AI training data, according to Rolling Out. Regulators are active in the wider AI market too. The Federal Trade Commission has sent civil investigative demands to OpenAI and Anthropic, as USA 24 reported.

Why publishers say the risk is not fully resolved

Sam Quigley, spokesperson for the News/Media Alliance, which represents US publishers, told the Mercury News: “It’s still potentially on the table.” The group says training was dropped from the listed uses but not expressly banned. The terms also do not promise that Meta will honor robots.txt, the standard file sites use to tell crawlers which pages they may visit.

That gap is the core of the dispute. Robots.txt only works when a crawler chooses to respect it. Meta says site owners can still set preferences there. Publishers want that commitment written into the contract, so it can be enforced rather than assumed. Until the company agrees to that, a publisher’s robots.txt file is a request, not a guarantee.

Which businesses should act before October 30

The steps below apply to news publishers and to any business with a Meta business account or Facebook page, since the terms reach websites associated with those accounts.

  • Read the revised commercial terms on Meta’s site before October 30. Check the listed purposes and the acceptance language.
  • Audit your robots.txt file. Confirm which pages and directories you block, then test the file after every change.
  • If you rely on Meta’s tools, ask Meta in writing whether AI training is excluded from crawling your site. Continued use counts as acceptance, so silence is not a safe default.
  • If your site carries licensed, paywalled or syndicated content, talk to counsel before the effective date.
  • Save copies of the terms you reviewed and any written replies from Meta, with dates.

What to watch before the terms take effect

Three questions remain open. First, will Meta explicitly rule out AI training or commit to honoring robots.txt? Publishers say neither appears in the revised text. Second, will lawmakers, regulators or publishers take further action before October 30? Nothing in the coverage confirms that yet. Third, reports differ slightly on the timing of the edit, with some dating it to October 16 and others to October 17.

The next two weeks will show whether Meta treats the revised wording as its final answer. An explicit training exclusion or a written robots.txt commitment would narrow the dispute quickly. If the terms take effect unchanged, expect publishers to press for clearer language, and expect the phrase develop, improve and research to become the test case.