Meta plans to implement a new version of its Commercial Terms of Service (TOS) starting October 30, 2026, requiring all business accounts to grant Meta default permission to use all content from external websites linked to by posts on Meta platforms such as Facebook and Instagram for training AI models. This clause does not offer an opt-out option; it merely advises businesses that do not wish to comply to stop using Meta’s commercial services.
This update affects all Meta business accounts, but it is particularly critical for news publishers: if their articles are continuously scraped for training purposes, Meta could use this to sidestep millions of dollars in content licensing fees. Currently, several news organizations are litigating whether OpenAI’s unauthorized use of their content to train AI constitutes fair use, while Meta’s new terms seek to pre-emptively establish broader data usage rights through contractual agreements—regardless of how relevant court rulings ultimately decide.
Ars Technica, citing Nieman Lab, reports that this provision effectively extends the “link is license” principle to AI training scenarios. Legal experts widely agree that this move poses significant compliance risks, particularly given the lack of precedent supporting its validity and enforceability under the EU’s Digital Services Act (DSA) and U.S. copyright law frameworks. The clause fails to specify key constraints such as the duration of data use, storage methods, or model types.