The interesting wrinkle is that this is effectively a consent laundering scheme. Individual users can’t meaningfully opt out of AI training across the ecosystem, and now business accounts ‘consent’ on behalf of every visitor whose data passes through their sites. The website operator agrees to terms covering content they don’t solely own — comments, bylines, photos of third parties.

Legally the news organizations are in a bind that’s worth watching: block Meta’s crawler and lose distribution, or allow it and hand over training data to a company building the models that flatten their traffic. The Nieman Lab angle makes sense — for publishers this is choosing between two slow bleeds.