Meta Drops Plan to Force AI Training on Linked Content
AI Policy·November 8, 2026

Meta has abandoned a planned update to its commercial terms that would have required Facebook Page owners to let the company use linked content for AI training. The company says it won't force Page owners to allow that use, which removes the most contentious part of the proposed change.
The clause raised concerns because a link is not a license. Pages routinely point visitors to articles, videos, product listings and other material that the page owner did not create. Under the proposed wording, a Page administrator could have been agreeing to AI training on work that belongs to someone else. Publishers and creators have been pressing AI companies for clear permission before their work is used to train models, and a platform quietly extending that reach would have drawn the same complaints.
The headline's question, "Does Meta think we can't read?", captures the frustration that such terms tend to provoke. Users are rarely fooled by vague consent language, and a change that depends on people skimming it tends to become a story in its own right.
The available reporting does not spell out the exact wording Meta has now settled on, or whether the company plans to revisit AI training permissions in a different form later. Page administrators who received notice of the update should review their current terms rather than assume the old ones still apply in full.
The episode is a small but useful signal. As platforms fold AI training into their policies, they are finding that consent language gets scrutiny, and that a clause granting broad rights over third-party material is hard to defend in public. Meta's reversal suggests that pushback can change the terms, at least this time.
Reporting based on an external source.