Meta removes business-terms clause that raised AI-training concerns

Meta has removed a commercial-terms provision that raised concerns about permission to scrape publishers’ websites and train AI models. The Verge reported the removal, according to TechCrunch’s October 17 report. The dispute centered on language covering content outside Meta’s platforms: websites belonging to businesses that hold accounts with the company.

Website content, not just account posts

The disputed wording said Meta could use content from a business account’s “domains, web pages, or URLs” to provide, develop, improve, research, train and maintain the integrity of its products and services. Its scope therefore extended beyond material a business posted directly to a Meta account. The list of permitted purposes included both operating products and training them, rather than describing website access alone.

Legal experts interviewed by Nieman Lab saw a potentially significant consequence for publishers with Meta business accounts. They interpreted the provision as possible permission for Meta to scrape those publishers’ websites—collecting their content through automated tools—and use the material to train AI models. That was an interpretation of the rights the wording appeared to grant, not a finding that Meta had begun training AI on publishers’ content under the clause.

We’ve heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we’ve clarified that in the terms themselves.

Meta, in a statement quoted by TechCrunch

A terms change, not a crawling halt

Meta’s statement also said: “It does not change how Meta uses web crawlers.” The company is presenting the revision as a clarification of its written terms, not a reversal of a newly introduced collection practice. That distinction limits what the removal means: Meta has changed the disputed language, but its response offers no promise to stop crawling publishers’ websites or to end AI training on web content.