Meta's outreach team uses AI and reviewers to flag potential school shooter threats

The team’s work draws on research into the so‑called True Crime Community (T.C.C.), an online fandom that glorifies mass‑shooters. The Institute for Strategic Dialogue reports that in the past two years the T.C.C. has been linked to at least nine attacks and sixteen disrupted plots. Meta says its automated review flagged content “around eight hundred times worldwide” last year, passing information to law‑enforcement in cases such as a Cardozo High School Instagram post that resulted in a gun seizure within two hours. The article also notes similar challenges at OpenAI, where a flagged account led to a tragic shooting after the company chose not to alert authorities.

Experts caution that while AI can help identify “leakage” of violent intent, platforms must balance privacy, liability and the risk of over‑reporting, especially as extremist language becomes more coded and harder for moderation tools to catch.