Meta’s Revised Business Terms Grant Sweeping Web-Crawling Rights for AI

Meta has rewritten the commercial terms that bind every business, brand and publisher on its platforms, and the new wording reaches far beyond Facebook and Instagram feeds. From 30 October 2026, companies that advertise, sell or simply run a business Page will authorise Meta to send automated crawlers across the domains and pages “associated with” their products — and to reuse whatever those crawlers collect in order to develop and train its own services.

• The scope is unusually wide: the licence names web scrapers, bots, spiders, crawlers and “user agents”, and adds the creation of test credentials to get past logins.
• There is no opt-out: carrying on with Meta’s business tools after the deadline counts as acceptance; the only alternative the contract offers is to stop using them.
• Meta denies any substantive change: spokespeople describe the text as a consolidation and say robots.txt still governs crawling.
• The clause has since been amended: after publisher backlash, Meta scoped the language to advertising and business products and removed the word “train”.

What the licence actually permits

Inside the “License & Rights You Provide” section, the clause authorises Meta “or any entity acting on its behalf” to deploy “automated software (such as web scrapers, bots, robots, spiders, crawlers, and user agents)” and to create test credentials in order to “access, index, cache, analyze, collect, retrieve, or crawl” the domains, pages and URLs tied to a company’s products or services — including links that arrive through users’ activity — and to pull periodic updates from them. A second paragraph extends this to “any content” from those sources, which Meta may use to “provide, develop, improve, research, train, and maintain the integrity of” its products. The clause never says “artificial intelligence”, but AI at Meta appears on the company’s official product roster.

Who is caught, and why refusing is impractical

The definitions sweep in advertising and marketing messages, product sales, app development, catalogues, Page and group administration, Ads Manager, Meta Business Suite and measurement services — “regardless of the type of entity”. Almost every newsroom qualifies. The notice appeared in Meta Business Suite on 2 October, flagged first by Stephen Beale, founder of The Steampunk Explorer, who has said he is weighing closing his Meta accounts and called the move “arrogant”.

The exit is theoretical: the terms state that anyone who disagrees “may stop all access or use of our Products for business or commercial purposes”. European firms face an extra wrinkle, since disputes with Meta Platforms Ireland are heard in Irish courts under Irish law unless the US parent is sued too. The clause is also a clear expansion of earlier text — the self-serve advertising terms in force since 22 August 2025 already allowed crawling for “research and training”, but only for companies buying ads, while the commercial terms of March 2024 contained no scraping language at all.

Meta’s defence, and the rewrite it triggered

Meta’s first response was flat denial. Spokesperson Daniel Roberts said the company was “consolidating” its commercial and self-serve ad terms and that “there is no change to how Meta uses web crawlers”. Andy Stone, vice president of communications, insisted the update was about product information for recommendations.

“This update is designed to make sure we’re able to pull accurate information about products for sale on third-party websites when making on-platform ad and content recommendations, not about scraping news sites,” Stone wrote.

Robots.txt was presented as the safety valve, yet Meta’s own crawler documentation concedes that one agent, Meta-ExternalFetcher, “may bypass robots.txt rules”; the agent used to train foundation models is a separate one, Meta-ExternalAgent. Under pressure, Meta then revised the text, saying it had “heard you” and that the crawling terms “only applied to our ads and business products” — a clarification now written into the contract, with “train” stripped from the disputed section.

The legal reading: a licence nobody negotiated

Researchers and practitioners see a licensing market being bypassed. James Grimmelmann of Cornell Law School put it bluntly.

“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.”

Matt Topic of Loevy & Loevy called it a remarkable provision to bury in terms of service, and estimated that the same access would cost “many millions of dollars” on the licensing market; Benjamin Sobel of the University of Wisconsin frames it as an unlawful tying arrangement, since crawling must be accepted to keep buying ads.

The strategic timing is what unsettles publishers. Meta has signed paid AI content deals with major outlets, has spent years training on public European user posts, and has simultaneously begun charging businesses and creators to include links in posts and reels. Granting itself free crawl rights over the destinations of those links inverts the economics of the open web. No evidence has surfaced that Meta is already using this clause against newsrooms, but from 30 October the permission exists in writing — and the fact that it had to be rewritten at all suggests the original drafting was never the harmless tidy-up Meta claimed.