Meta deleted language that appeared to permit website content collection for AI training, while saying the revision does not change its crawling practices.
On October 17, Meta removed wording from its commercial terms that had said it could use content from websites associated with business accounts to develop, research, and train its products. The change followed criticism from publishers and legal experts, who warned the earlier language could be read as permission to collect material from those sites for AI training.
The deleted provision covered content on domains, webpages, and URLs linked to a business account. Meta’s statement to The Verge said the crawling terms applied only to its advertising and business products and that the company had clarified this in the terms. Meta also said the revision does not change how it uses web crawlers, and that site owners can continue to use robots.txt to signal access preferences.
The dispute concerns more than posts shared directly on Facebook or Instagram: the earlier wording addressed material on associated external websites. Meta’s updated terms were scheduled to take effect on October 30, making the wording relevant to businesses that use its commercial products. The revised language removes the explicit reference to training, but Meta’s explanation does not amount to a promise to stop crawling websites.
For publishers and other businesses, the episode highlights how contract language can shape expectations about web content and AI development. Before the revision, legal specialists told Nieman Lab that the provision could be interpreted broadly; Meta rejected the suggestion that its crawling practices had changed. The updated wording narrows the stated purposes, while leaving the company’s description of its crawler behavior unchanged. The practical effect will depend on the final terms and how Meta applies them after they take effect.