The central dispute concerns the word "train." An earlier version of the clause listed AI training among permitted uses. After objections from lawyers and publishers, Meta removed that word. Critics say the revision neither bans AI training outright nor commits Meta to honoring robots.txt files. Three legal experts who reviewed the text for Nieman Lab said it could still be read to authorize training Meta's AI on scraped content. The timeline is also unresolved. One source says the date was pushed from October 30 to November 20, 2026, and Meta has said the terms are delayed and not in effect. Other reports still list October 30 as the effective date. The current status has not been confirmed.
This matters because the effective date is close and the scope extends well beyond news publishers. Any business with a Facebook Page or Instagram profile may be covered, and the clause could reach every domain linked to a business account. The News/Media Alliance called the terms "outrageous and probably illegal," and law professor James Grimmelmann said publishers must either accept Meta's narrower reading or stop using business accounts. Attorneys advising clients with business accounts should identify which domains are linked, review their robots.txt configurations, and monitor Meta for the confirmed effective date and any further revisions.