Meta flagged around 800 violent threats to police last year

Meta reported online content to law enforcement roughly 800 times last year to try to head off violent attacks worldwide, including school shootings, according to a New Yorker piece by Rachel Monroe profiling the company's law enforcement outreach team.

What the team flags is wider than public posts. The referrals, Monroe reports, include "posts, interactions, and prompts fed to A.I. chatbots." Conversations users have with Meta's own AI products feed into what the team watches, alongside social-media content. The workflow pairs AI classifiers with human reviewers who decide what gets passed on to police.

The magazine's framing is a reversal: algorithmic social media, Monroe writes, "which has helped create the conditions for the spread of such violence, is reckoning with how it can be part of the solution." The 800-report figure is Meta's own count, as given to the magazine, with no independent audit alongside it. The piece lands in the middle of a long run of safety coverage we have been tracking over the last ninety days.