Meta updates terms to crawl business sites for AI
Meta updated its commercial terms for October 30th to expand crawler access across business accounts, potentially letting it train AI models on publisher content without paying licensing fees.
Meta is revising its commercial and self-serve advertising agreements effective October 30th, giving itself explicit permission to use web crawlers, scrapers, and automated bots to index and collect material from domains tied to business users. The policy applies to commercial accounts active across Facebook, Instagram, WhatsApp, and Messenger. Under the revised conditions, Meta may utilize content pulled from these linked websites to develop, research, improve, and train its artificial intelligence models and platform features.
Legal analysts, including Cornell law professor James Grimmelmann, warn that the clause could allow Meta to argue its commercial agreement supersedes standard web directives like a site's robots.txt file. Though Meta has secured formal AI training deals with publishers like CNN, Fox News, USA Today, and People Inc., these terms potentially grant the company free access to a much wider array of web content. In response to reporting from Nieman Lab, Meta spokesperson Daniel Roberts asserted that the update merely consolidates existing terms and that developer documentation regarding robots.txt remains unchanged. Additionally, Meta communications vice president Andy Stone posted on X that the change aims to improve product data accuracy for ad recommendations rather than target news publishers.
For enterprise practitioners, publishers, and social media strategists, the policy presents a mandatory compliance choice. Organisations unwilling to accept the terms must stop using Meta's products for commercial goals. The decision is further complicated by Meta's launch of its Meta One subscription, which requires businesses to pay for including links in reels and posts, effectively requiring organizations to pay for platform reach while Meta potentially scrapes their external websites for free.
Practitioners must carefully assess their data policies and web crawler settings ahead of the deadline to balance social engagement against uncompensated AI model training on their proprietary content.
This is our own summary of reporting by The Verge AI
