Deleting one word is not the same as giving up a right. Meta has tweaked its commercial terms of service after publishers objected, and it removed the reference to AI training. But the core grant is intact: Meta can still send automated tools to crawl the websites of anyone with a business account on Facebook or Instagram.
Here is what happened. Meta announced an update to its commercial terms last week, due to take effect October 30. As Nieman Lab flagged, the language said Meta could use scrapers and other automated tools to “access, index, cache, analyze, collect, retrieve, or crawl” the domains, pages and URLs tied to business accounts. It also said Meta could use that content to “train” its products. Then Meta quietly revised the text. “Train” is gone, and scraping is now described as serving to “provide, develop, improve, research, and maintain advertising and business products.”
Meta spokesperson Andy Stone said the language was “clarified” to show the terms apply “only to our ads and business products,” and that the changes “were not about scraping news sites.” He also said nothing changes about how Meta uses web crawlers.
That’s a tidy answer, but it skips the problem. Legal experts told Nieman Lab the revised terms are still a substantial expansion of Meta’s scraping permissions, and that Meta can scrape news sites with few apparent limits. “Advertising and business products” is also a wide category, and the terms define it broadly, including products that help businesses create ads and connect with users. Who decides where that ends? Meta does.
Stone also pointed to robots.txt, saying the update doesn’t give Meta new rights to override publishers’ preferences. But robots.txt is a courtesy, not a legal barrier. It works only if the crawler chooses to respect it, and AI companies regularly don’t. A contract clause that authorizes crawling is a different thing from a polite request to stay out, and the contract is what you would be pointing to in a dispute.
For users, the stakes go beyond publishers. Anything on a linked domain, from storefronts to community pages, could be pulled into Meta’s systems, and the stated purposes are vague enough to stretch. Once content is cached and analyzed, a narrower wording on paper does little to show where it ends up.
The history doesn’t help Meta’s case. It stopped paying for news in 2022, shut down its news tab in 2024, and has more recently signed a handful of deals with outlets to supply news for its AI chatbot. Publishers have reason to read generous terms as leverage.
They are reading it that way. Danielle Coffey of the News/Media Alliance called the rules “outrageous and probably illegal.” Theo Bamber of the UK’s News Media Association told Press Gazette it was a “shameless landgrab attempt.”
The takeaway: judge Meta by what the contract permits, not by what a spokesperson says the company currently does. Terms outlast statements.