Meta withdraws clause on training AI on company website content
After a wave of criticism from lawyers and publishers, Meta revised the terms for business accounts that take effect on October 30. Some analyses indicate, however, that the right to collect content from company websites remains in the terms.
Meta revised the updated commercial terms for business accounts. It removed a clause under which it could use content from domains, websites and URLs associated with those accounts, among other things, to develop, research and train its products and services. The Verge reported the withdrawal of the clause on October 17, and TechCrunch cited its findings. The new terms are to apply from October 30, 2026.
The issue was publicized earlier, around October 16, by Nieman Lab. Lawyers interviewed by the outlet said the clause could have given Meta permission to scrape, that is, automatically collect content, from the websites of every publisher with a business account. The collected content could have been used to train AI models. The terms cover companies that run pages or profiles, advertising campaigns, or marketing message distribution on Facebook, Instagram, WhatsApp and Messenger, as well as companies using the Meta Business Suite dashboard. According to Nieman Lab, the terms did not provide an opt-out option.
Meta argues that the change does not affect the operation of its crawlers, that is, programs that automatically browse websites. In a statement cited by TechCrunch, the company explains that the crawling provisions, previously scattered across several documents, were merely collected in one place. According to Meta, they apply only to advertising and business products.
Sources differ on the scope of the revision. TechCrunch writes that the entire clause was removed. Some analyses say that only the word “train” disappeared from the list of permitted uses, while the remaining uses were tied to advertising and business tools. The authorization to crawl domains associated with business accounts was reportedly left in the terms. Meta says that website owners can restrict access with a robots.txt file, that is, a file with instructions for indexing bots. According to Nieman Lab, however, the terms do not contain a commitment that the company will respect those instructions.
For companies with a Meta business account, this means that the clause on the purpose of using content has changed. According to some analyses, consent for the collection of that content itself has not been withdrawn. For companies, the basic control tool remains the robots.txt file on their own website.