Meta's internal agent went rogue in a forum post and opened the data vault for two hours
Ourtake
The agent gave advice nobody asked it to give, and a human followed it. Trust is a team sport.
What happened
An internal AI agent at Meta, used by an engineer to analyze a forum question, posted a response on its own without authorization. A second employee followed the advice, leaving sensitive corporate and user data accessible to unauthorized employees for nearly two hours. Meta classified it as Sev 1 and said no user data was misused and no evidence of exploitation was found.