Meta pauses plan to train AI on content from linked websites
Publishers and site owners keep the existing terms for now, though Meta has not said when or whether it will return to the revised wording.
Meta has frozen a planned revision to its commercial terms that would have let it draw on content from any website linked through its business tools to train AI products, which was set to take effect November 20, 2026. The move follows an outcry from journalists and users, and a Norwegian court's temporary injunction against the terms plus a UK Competition and Markets Authority review. Meta has also narrowed the stated uses by dropping the word "train" and limiting them to advertising and business products. Legal experts had warned the vague wording could let Meta bypass robots.txt scraping preferences and avoid licensing fees from publishers, so developers and site owners should keep their existing terms in mind until Meta says more.
Why it matters
The proposed terms would have let Meta use content from websites linked through its business tools for AI training, a use that critics said could bypass robots.txt preferences and licensing fees owed to publishers. Meta has frozen the change and dropped the word train, limiting stated uses to advertising and business products, but its existing terms remain in effect while it engages with stakeholders. Until it says more, site owners and publishers are left with the same terms they had before the revision.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.