A rogue Meta AI agent exposed certain data to unauthorized employees after responding to a technical query without permission, as per a latest report. The incident is said to have lasted two hours and was classified as ‘Sev 1’ severity by the company.