Meta has revised its terms of use following the debate over AI training. The wording has been removed from the text. While the company dropped the phrase stating that content on websites linked to business accounts could be used to “train” its products, the previous text had raised concerns among publishers and legal experts about AI training and copyright. In the new terms, which take effect on October 30, Meta retains the right to scan these sites with automated tools and states that the aim is to improve advertising and business products. The company maintains that there has been no change in practice. Spokesperson Daniel Roberts said site owners could set their preferences using robots.txt. Andy Stone, for his part, explained that the arrangement does not target AI training. Since the scanning right remains in place, it is hard to say the debate is over.
Why it matters
The change is the latest example of the ongoing debate over how AI companies’ access to content on the internet should be regulated. Although the removal of the wording is a step toward easing publishers’ copyright concerns, the retention of the scanning right leaves open the question of for what purpose this content will be used in the future. Those directly affected by the change are site owners and content creators who connect to Meta platforms through business accounts; for them, the only practical safeguard is the option the company points to of declaring preferences via robots.txt. However, it is unclear how much protection this mechanism alone will provide. The gap between the company’s claim that there has been no change in practice and the concerns of legal experts raises questions about how much changes in contract texts reflect actual policy.
Background
Meta is not a new name in the FikirPilot archive: we have published 41 news stories mentioning this name in the last 90 days; the most recent is dated October 13, 2026.