Meta commercial terms may allow AI training on linked sites

Meta’s October 30, 2026 commercial terms would let it crawl linked business sites to train AI, with licensing and opt-out effects disputed.

Meta’s updated commercial terms, effective October 30, 2026, authorize crawlers, bots, and test credentials to access domains associated with business accounts on Facebook, Instagram, WhatsApp, and Messenger, and allow that content to be used to develop, research, and train Meta AI products. The Verge, citing Nieman Lab, says the language could cover online publishers and bypass separate licensing deals with outlets including CNN, Fox News, USA Today, and People Inc.; businesses that reject the terms may lose commercial access, and Meta One already charges some users to post links. Ars Technica reports that business accounts would be required to authorize the crawling with no opt-out, and that legal experts told Nieman Lab the clause could support training on publisher sites while avoiding licensing fees worth millions. Meta told Ars the change consolidates existing crawling language, that robots.txt is still honored, and that sites can block crawlers. Ars says wording such as “test credentials” did not appear in the prior commercial, catalog, or ad-creative AI terms. The reports agree on the effective date and AI-training use but differ on whether this is consolidation or a route around paid licenses.

• Commercial terms take effect October 30, 2026.
• The terms authorize crawlers, bots, and test credentials to access domains tied to business accounts on Facebook, Instagram, WhatsApp, and Messenger.
• Meta may use that outside content to develop, improve, research, and train its AI products.
• Nieman Lab and legal experts say the language could cover publisher sites and undercut licensing deals with outlets including CNN, Fox News, USA Today, and People Inc., potentially avoiding fees worth millions.
• The Verge says businesses that reject the terms may lose commercial access; Ars Technica says there is no business opt-out.
• Meta says the update consolidates existing crawling language, that it still honors robots.txt, and that sites can block crawlers.
• Ars Technica says “test credentials” wording is new and was absent from prior commercial, catalog, and ad-creative AI terms.
• Meta One already charges some users to post links.