Meta’s new business terms spark AI scraping fears

Meta is updating the commercial terms that cover business accounts on its platforms, and news publishers are worried. The new terms take effect on 30 October and, as first reported by Nieman Lab and analysed by Ars Technica, they give Meta broad permission to crawl the websites linked to those accounts.

What the terms say

Businesses that accept the terms authorise Meta to use automated software such as scrapers, bots and crawlers, and to create test credentials, in order to access, index, cache and collect content from domains associated with their products or services. Meta may then use that content to “provide, develop, improve, research, train, and maintain” its products.

The text never mentions AI directly. Concern centres on the word “train” and the scraping language. There is no opt-out: a business that disagrees would have to stop using Meta’s services. Since virtually every news organisation runs at least one Meta business account, Nieman Lab says publishers are particularly exposed.

Meta’s response

Meta calls the change “streamlining”. A spokesperson said crawling language was previously scattered across different terms, including those for generative AI in ads, and has now been consolidated. The company says it has no plans to use the terms to ignore robots.txt, still lets news organisations block its crawlers, and told Ars that “there is no change to how Meta uses web crawlers.”

Why experts are uneasy

Three legal experts told Nieman Lab it’s unclear whether a court would enforce such terms, but that they could be used to argue Meta may scrape news sites and train AI on that content. Cornell law professor James Grimmelmann warned Meta could even deploy some of that content in competition with publishers. Experts also flagged the “test credentials” wording, which they fear could help get past paywalls.

Ars notes that this phrase doesn’t appear in Meta’s current commercial terms, though Meta pointed to a Seller Agreement that uses similar language for a narrower e-commerce purpose. Media lawyer Matt Topic called it “a remarkable provision to bury” in the terms, given that such access could otherwise cost “many millions of dollars” in licensing.

For more on Meta’s AI push, read about its Muse Wonder models. If you’d rather keep AI on your own machine, see our guide to choosing a laptop for local AI.