Meta restricts news scraping in AI training terms after publisher pushback

Meta's reversal happened within 24 hours of the original policy going live, indicating the removal was reactive to publisher backlash rather than proactive governance. The critical omission: the summary doesn't clarify whether this blocks scraping entirely or merely requires explicit consent, a distinction that determines whether the policy actually restricts Meta's data acquisition or simply shifts it to a negotiation phase.

This directly contradicts yesterday's Verge story about Meta's new terms granting broad scraping rights. The one-day cycle reveals how quickly publisher pressure forces reversals, but it also exposes the underlying dynamic from the dead-startup data acquisition piece (mid-October): when training data becomes scarce, companies test boundaries through policy language, then retreat when caught. The pattern suggests contractual guardrails are performative unless paired with enforcement mechanisms publishers actually control.

If Meta negotiates paid licensing deals with major publishers within 60 days, the policy shift was genuine. If scraping resumes under different contractual framing (opt-in rather than opt-out), or if Meta pivots to acquiring training data through alternative channels like the startup liquidation market, the removal was theater. Monitor whether other labs follow with similar reversals or double down on existing scraping practices.