Meta has removed language from its commercial terms that critics said could let the company scrape linked websites, including news publications, to train AI systems. The deleted passage said Meta could use content from domains, pages or URLs to develop, improve, research and train its products.
Legal and media observers interpreted the wording as an attempt to secure broad rights over content linked through Meta’s business products. Some also questioned whether it could override a publisher’s instructions in robots.txt, the standard file websites use to signal which automated crawlers may enter.
Meta disputed that interpretation. Spokespeople said the consolidated terms applied only to ads and business products, did not change how its crawlers operate and did not grant rights that supersede site owners’ preferences. The company has now removed the contested language and added a clarification after the backlash.
The edit narrows what the document appears to authorize, but it does not independently establish what content Meta has collected or used for training. Publishers must still rely on the company’s stated crawler policy, technical blocking tools and any applicable legal rights. The episode shows how a broadly written platform contract can create uncertainty even when a company says its underlying practice has not changed.