Meta’s New Commercial Terms Could Expand AI Training Access to Business Websites

Summary

Meta is preparing commercial terms that will take effect on October 30 and require business accounts to authorize automated software to access, index, cache, analyze, collect, retrieve, and periodically crawl websites or URLs associated with their products and services. The terms say Meta may use content from those sites to provide, develop, improve, research, train, and maintain its products, although they do not explicitly mention AI. Businesses would not receive an opt-out under the proposed arrangement; Meta’s position is that companies unwilling to accept the terms should stop using its services. The change is especially consequential for news publishers, which are already contesting with OpenAI whether AI training on articles without licensing is fair use and could otherwise seek millions of dollars in licensing fees. Meta says it is merely consolidating crawling language previously scattered across several policies, that it has no plan to override robots.txt, and that content owners can still block crawlers. Ars Technica and legal experts point to broader wording in the new terms, including references to “test credentials,” which could potentially enable automated access behind paywalls. Earlier policies cited by Meta limited crawling to specific purposes such as ad generation or e-commerce listing optimization, while the new commercial terms apply more broadly to Meta products and services. Experts told Nieman Lab that a court’s view of enforceability is uncertain, but the terms could be used to argue that Meta has permission to train AI on publishers’ content and potentially deploy competing uses of that material. They also compared the arrangement to tying, because access to Meta’s business services may be conditioned on granting additional content permissions. The dispute is developing alongside Meta’s new charges for posting links, prompting some publishers and at least one platform-dependent website operator to consider reducing or ending their use of Meta services.