Starting October 30, Meta’s new commercial terms shift liability for any AI errors onto enterprise customers and hold IT executives accountable for partners’ actions, while expanding Meta’s rights to scrape and use customer-related content, raising data privacy concerns. These changes have been criticized for unfairly burdening customers with risks they cannot control and undermining traditional data protections, potentially forcing businesses to reconsider their IT and data management strategies.
Meta is implementing new commercial terms starting October 30 that shift liability for any mistakes made by its AI models onto its enterprise customers. This means that if Meta’s AI produces inaccurate or unsuitable results, the customer bears all associated risks. Additionally, the terms hold IT executives responsible for actions taken by partners or contractors, effectively transferring accountability away from Meta despite the company acknowledging that its AI may err.
A significant change in the terms involves Meta’s expanded rights to web scraping and data ownership. Meta now claims the right to use automated tools like web scrapers and bots to access, index, and analyze content associated with Meta accounts, including content uploaded or shared by customers or their agencies. Although Meta later clarified that these crawling terms apply only to its advertising and business products and do not override robots.txt preferences, the initial language raised concerns about data control and privacy.
The AI liability clause has drawn criticism from industry experts who argue that it unfairly places the burden of AI errors on customers, despite Meta controlling the AI models. This shift could deter enterprise customers wary of being held responsible for outputs they cannot directly control. The timing of the changes, just before Halloween, was noted ironically by analysts as fitting for a move seen as a “trick” rather than a “treat.”
Meta’s new terms also grant the company broad licenses to use customer content for developing and maintaining its products, with licenses surviving even after contracts end. This expansive language, combined with the removal of explicit references to AI training on enterprise data, has been described as legal sophistry—altering wording without changing the substantive rights Meta claims over customer data and content.
These changes may force enterprises to rethink their IT strategies, especially since traditional protections like robots.txt are now effectively undermined by contractual terms. The broad scope of Meta’s rights extends to content uploaded by agencies and vendors and to domains loosely associated with customers’ brands, complicating oversight and control. As a result, businesses may face challenges in managing their digital presence and data privacy in dealings with Meta’s platforms.