Meta shifts AI error liability to enterprise customers

Meta's new commercial terms, effective October 30, make enterprise customers solely responsible for mistakes made by its AI features and for actions of partners. The company also grants itself broad rights to scrape and use content tied to Meta products, overriding robots.txt preferences.

• New terms take effect on October 30, 2026
• Customers assume "any and all risks" for inaccurate AI outputs
• Meta allows scraping of domains and pages linked to its products
• Content license survives termination of the agreement