Meta's New Business Terms Could Let It Train AI on News Sites for Free

Meta has rewritten its Commercial Terms for businesses, and publishers are treating the update as one of the most consequential changes to the platform's relationship with the press in years. According to reporting by The Verge, drawing on legal analysis first published by Nieman Lab, the revised terms — which take effect on 30 October 2026 — give Meta permission to crawl the websites associated with business users and to use the collected content to train its artificial-intelligence products. Because almost every news organisation operates at least one Meta business account, the clause effectively reaches the entire industry's output.

The change was not announced with fanfare. Meta disclosed it in a blog post in early October 2026, and Nieman Lab learned of it only after a tip from Stephen Beale, founder of The Steampunk Explorer, who spotted the new language in his Meta Business Suite dashboard. What he found was a clause that allows Meta to deploy automated software against the domains, pages and URLs tied to a business's products or services — and to return periodically for updates.

What the new terms actually say

The core of the dispute is the breadth of the new clause. The terms authorise Meta — or an entity acting on its behalf — to use scrapers, bots, robots, spiders, crawlers and user agents to access, index, cache, analyse, collect, retrieve or crawl the websites connected to a business user, and to pull fresh copies on a recurring basis. Once collected, the content may be used to "provide, develop, improve, research, train, and maintain the integrity of" Meta products — and Meta's own terms define "AI at Meta" as one of those products.

That last phrase does the heavy lifting. Earlier Meta language, in the Ad Creative Generative AI Terms effective from May 2024, already permitted automated crawling of URLs that users supplied in connection with advertising, but it was framed around ad use. The new Commercial Terms replace that narrower framing with a much broader business-associated-site trigger, and training is named explicitly among the permitted purposes.

For news publishers, the practical consequence is stark. As Cornell law professor James Grimmelmann put it: "For news publishers, this clearly represents permission to Meta to train AI on any content from their sites and possibly to deploy some of that content, even in competition with the publishers themselves."

Why there is no way to say no

The terms contain no opt-out. The only remedy Meta offers dissenters is exit: "If you do not agree to the updated terms, please stop all access or use of our Products for business or commercial purposes." In other words, a publisher that wants to keep reaching audiences through Facebook, Instagram, WhatsApp channels or Messenger — or to keep running advertising through Ads Manager — must accept the scraping permission as a condition of staying on the platform.

Three legal experts reviewed the terms at Nieman Lab's request, and the verdict from media and intellectual-property attorney Matt Topic was blunt: "It's not clear whether a court would enforce this, but publishers should assume that it would be enforceable."

University of Wisconsin–Madison law professor Benjamin Sobel compared the structure to "tying" in antitrust law: to use the Meta products that require the commercial terms, businesses must also accept the scraping permission. Sobel drew a parallel with Google's policy tying publishers' appearance in search to the inclusion of their content in AI Overviews and AI Mode — another case where platform participation comes bundled with AI-training consent. The early publisher reaction has been alarm: Beale told Nieman Lab he is seriously considering shutting down his accounts before 30 October, describing the move as "an arrogant, in-your-face move".

Robots.txt, paywalls and the supersession question

Meta's public line is reassuring on the surface. A public-affairs spokesperson said by email that the update merely consolidates crawling language previously scattered across different sets of terms, that it will not change how Meta crawls publishers, and that content owners can still manage crawler access through robots.txt. Legal experts, however, caution that the contractual wording could be used to argue it supersedes robots.txt opt-outs — a site owner that agreed to the Commercial Terms may have signed away the very protection the robots file was supposed to provide.

The clause also permits Meta to "create test credentials" to access associated sites. Grimmelmann said that language would technically allow Meta to "sign up as a fake user" to get past registration walls, though Meta declined to say whether the permission extends to paywalls or paid content. The open question matters: a crawler that can register an account is a crawler that registration walls were never designed to stop. This remains one of the follow-ups the industry will be watching closely once the terms bite.

A fight already running in the courts

The terms arrive against a legal backdrop that is anything but settled. Several lawsuits between publishers and AI developers over training data and fair use are working their way through federal courts, including a class action filed in May by five large book publishers alleging Meta pirated millions of books to train its Llama models. Against that background, Meta declined to answer a detailed list of questions from Nieman Lab — and has not confirmed whether it actually intends to begin scraping publisher sites under the new terms, or to use scraped content for training.

The unresolved legal theory is the crux. Whether consent buried in platform terms of service can override a copyright fair-use analysis, whether it can override robots.txt, and whether the tying structure attracts antitrust or regulatory scrutiny — including under the EU's Digital Markets Act — are all questions no court has yet answered. Readers following this developing area of AI research and policy can expect the answers to shape how every major platform handles training data.

The broader squeeze on publishers

The terms change does not stand alone. It lands alongside two other Meta moves that tighten the economics for publishers: the paid subscription service Meta One, which charges for links and limits external links on Facebook and Instagram (Press Gazette reports some news publishers are currently exempt from the link limit), and the rollout of Muse, Meta's new agentic AI app. The combination has a particular sting, as observers have noted: Meta charges publishers for links to content that it may then crawl for free.

Whether that perception hardens into legal action or regulatory pressure after 30 October will define the next phase of the publisher-platform relationship. For now, the industry's choice is binary, and the deadline is fixed.

FAQs

What has Meta actually changed?

Meta rewrote its Commercial Terms, effective 30 October 2026, so that anyone using Meta products for business or commercial purposes agrees Meta may crawl the websites associated with their business and use collected content to "train" its products — which Meta defines as including its AI. Because virtually every news organisation uses at least one Meta business product, publishers' websites fall within the clause's reach.

Why can't publishers just opt out of the scraping clause?

The terms offer no opt-out mechanism. The stated alternative is to stop using Meta's products for business or commercial purposes entirely — meaning no advertising, no business Pages or Groups, no Business Suite, no commercial use of Instagram, WhatsApp or Messenger. Media attorney Matt Topic advises publishers to assume a court would enforce the clause.

Does robots.txt still protect a publisher's site?

Meta says yes: its spokesperson stated that content owners can still manage crawler access via robots.txt. But legal experts warn the new contractual consent could be argued to supersede a robots.txt opt-out in court, since the business user has explicitly agreed to the crawling. The conflict has not yet been tested before a judge.

Could Meta use the terms to bypass paywalls?

The terms let Meta "create test credentials" to access associated sites, which Cornell's James Grimmelmann says would technically allow Meta to "sign up as a fake user" to slip past registration walls. Meta declined to say whether the permission covers paywalls or paid content, so that boundary remains unconfirmed.

What should publishers watch for next?

Three things: whether Meta confirms it will actually scrape publisher sites under the terms and use the content for training; whether publishers bring lawsuits or regulators open scrutiny — particularly around tying and antitrust angles — after 30 October; and whether any court tests the argument that contractual consent overrides robots.txt. Any of those could reset the industry's terms of engagement with AI platforms.