Meta deletes AI-training language from commercial terms after legal experts flag publisher scraping risk

On October 17, 2026, Meta removed a provision from its updated commercial terms of use that legal experts said could let the company scrape content from websites tied to Meta business accounts — including news sites — and use it to train its AI models.

The original clause, first surfaced by Nieman Lab on October 15, said: "Meta may use any content from such domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of, the Meta Company Products, services, and features provided under these Commercial Terms and any other applicable supplemental terms." It applied to any business holding a Meta business account (e.g. a Facebook Page or Instagram business profile), with the updated terms taking effect October 30, 2026.

Three legal experts reviewed the wording for Nieman Lab. Media and IP attorney Matt Topic called it "a remarkable provision to bury in terms of service," saying it would give Meta access to material that would cost many millions of dollars in a licensing market. Cornell digital-law professor James Grimmelmann said the terms read as permission to train AI on publishers' site content and possibly deploy it in competition with the publishers. Critics also argued the language could override a site's robots.txt blocking instructions. The News/Media Alliance called the initial terms "outrageous," and Nieman Lab reported virtually every news organization holds at least one Meta business account, with no opt-out option ahead of October 30.

Meta deleted the clause on October 17. Spokesperson Daniel Roberts said the consolidated crawling terms "only applied to our ads and business products" and that the edit "does not change how Meta uses web crawlers." VP of communications Andy Stone said on October 16 that the update was about pulling accurate product listings for ad recommendations, "not about scraping news sites." The revised wording ties crawling to providing and maintaining "advertising and business products" and drops the verb "train."

Not all of the dispute is resolved: the permission to crawl domains associated with business accounts remains in the terms, including authorization to "create test credentials," and a UK English edition of the terms preview was found on October 19 to still carry the original "train" sentence. The News/Media Alliance said training is "still potentially on the table" since it was not expressly ruled out.