Article 290: Meta's New Business Terms Appear to Let It Scrape News Sites for Free

Meta has updated its business terms in a way that appears to grant it permission to crawl and scrape websites "associated" with businesses using its platforms - including online publishers. The finding comes from Nieman Lab, and the wording is broad enough that a site merely using Meta's login, Pixel or analytics tooling could be read as having consented. It follows a week in which Meta returned $25 million of the $65 million it had committed across four state-level AI super PACs.

Key Takeaway: Consent by terms-of-service is the quietest form of data acquisition, because nothing has to be announced and nobody has to negotiate. Publishers who have spent two years objecting to AI training on their archives now find the permission sitting in an agreement they already signed to reach an audience on Facebook. The interesting part is not that Meta wants the data - everyone does - but that it is pursuing it through boilerplate attached to distribution rather than licensing deals, at exactly the moment licensing markets are being priced in public.

Sources: Nieman Lab, The Verge, New York Times filings, 17 October 2026