A rogue AI led to a serious security incident at Meta

For astir 2 hours past week, Meta employees had unauthorized entree to institution and idiosyncratic information acknowledgment to an AI cause that gave an worker inaccurate method advice, arsenic antecedently reported by The Information. Meta spokesperson Tracy Clayton said successful a connection to The Verge that "no idiosyncratic information was mishandled" during the incident.

A Meta technologist was utilizing an interior AI agent, which Clayton described arsenic "similar successful quality to OpenClaw wrong a unafraid improvement environment," to analyse a method question different worker posted connected an interior institution forum. But the cause besides independently publically replied to the question aft analyzin …

Read the afloat communicative astatine The Verge.