Meta adds then removes AI scraping language from business terms

By Splitfeed EditorialLast updated Oct 17, 2026, 16:30 UTC

On or around October 16, 2026, Meta quietly updated its commercial terms to include language permitting it to crawl and scrape websites linked to business accounts on Facebook, Instagram, WhatsApp, and Messenger. The new wording said Meta could use content from those domains to "develop, improve, research, train, and maintain" its AI products, with the policy set to take effect October 30, 2026. Businesses continuing to use Meta platforms past that date would be considered to have accepted the terms. Nieman Lab flagged the policy first, reporting it could allow Meta to train AI models on news publisher websites without payment or explicit consent. By October 17, 2026, Meta had removed the scraping language from its commercial terms page. The company has not stated publicly whether any data was collected under the original wording before the removal.

The episode matters because it sits at the intersection of two already contentious debates: who controls the web content that trains large AI models, and whether platform terms of service can substitute for genuine consent. News publishers and other businesses with linked domains had no opportunity to negotiate the terms before they were posted. The opt-out mechanism was simply continuing to use the platforms, which for many businesses is not a realistic choice. The speed of the removal, less than 24 hours after Nieman Lab's report, suggests Meta judged the public and legal exposure to outweigh whatever data access the terms would have provided. What remains unresolved is whether the language will return in a revised form, and whether any scraping occurred in the window between the policy being posted and being pulled.

No reactions from the Pro-AI, Anti-AI, or Middle Ground camps have been published yet on this story. The Pro-AI camp would typically argue that training on publicly accessible web content is consistent with how AI development has always worked and that Meta responded appropriately to concerns by reversing course quickly. The Anti-AI camp would likely treat this as evidence that AI companies routinely attempt to expand data rights through obscure contractual language, counting on low visibility before rollout. The Middle Ground camp would probably call for clearer regulatory standards that distinguish between legitimate crawling and covert data extraction, so that businesses understand their rights before terms change.

The clearest signal to watch for is whether Meta reintroduces similar language in a future terms update, and whether any legal challenge or regulatory inquiry follows from the brief window the original wording was live.