Meta Revises Business Terms Over AI Scraping and Liability

Meta Platforms, Inc. revised its upcoming commercial terms of service after reports suggested the language allowed broad scraping of websites to train AI models. In updates published Friday, the company removed the word train from its allowed use cases and restricted the use of scraped content specifically to advertising and business products rather than all company products.

The updated terms, effective October 30, 2026, also shift legal liability for AI-generated errors to enterprise customers. Meta now states that customers are solely responsible for any risks associated with inaccurate or unsuitable results produced by AI features. This expansion of liability includes content provided by a customer's partners or contractors.

Public affairs manager Emil Vazques clarified that the crawling terms do not change how the company uses web crawlers and do not supersede robots.txt preferences. Meta stated the revisions were necessary because the initial consolidation of terms did not clearly indicate that these preferences are respected. Despite these clarifications, industry analysts warn that the terms create a one-way street regarding data ownership and may force enterprise customers to change their data management strategies.