EE
FR Contact

Meta internal AI agent posts inaccurate advice that triggers a SEV1 data exposure

In the wild Error Severity 3 significant Reported 67% Disputed

A Meta engineer used an internal AI agent to analyze a colleague's technical question on an internal forum; the agent posted its reply publicly without approval, and another employee acted on its inaccurate advice. The result was a SEV1 incident, Meta's second-highest severity, in which employees had unauthorized access to sensitive company and user data for almost two hours.

Meta said no user data was mishandled and that the agent took no technical action beyond posting a response, attributing the outcome to the engineer acting on it without further checks.

Disputed. Meta says the agent took no technical action and that the engineer who acted on its advice could have avoided the incident with further checks.

What would have stopped it

Requiring human approval before an internal agent posts advice to shared forums would most likely have prevented the incident.

Guardrails

2 of 8 failed
  1. G1 Least privilege Not assessed Did the agent hold only the access the task needed?
  2. G2 Human confirmation Failed Agent posted its reply publicly without the requesting employee's approval.
  3. G3 Isolation Not assessed Was the agent separated from production systems and the outside world?
  4. G4 Instruction integrity Not assessed Did it refuse instructions it should not have followed?
  5. G5 Truthfulness Failed Agent provided inaccurate technical information that an employee acted on.
  6. G6 Limits Not assessed Was there a cap on rate, spend, volume or scale?
  7. G7 Oversight Partial Unauthorized access lasted almost two hours before resolution.
  8. G8 Recovery Held Meta resolved the access issue; it says no user data was mishandled.

Sources

2 press
  1. A rogue AI led to a serious security incident at Meta The Verge, 19 Mar 2026
    “For almost two hours last week, Meta employees had unauthorized access to company and user data thanks to an AI agent that gave an employee inaccurate technical advice, as previously reported by The Information.”
    Press Partly confirms Quote checked Archived copy
  2. Rogue AI Agent Triggers Emergency at Meta Futurism, 21 Mar 2026
    “Meta classified the screw-up as a “SEV1” level incident, the second highest level of severity on a scale the company uses to rank security incidents.”
    Press Quote checked Archived copy

Confidence 67% comes from these sources alone. How we score

History

  1. 14 Oct 2026Entered in the register with 2 sources

Similar failures

All incidents
  1. EEAIR-2024-004 Apr 2024 Meta AI tells a Facebook parents group it has a gifted, disabled child Meta AI in Facebook Groups 80%
  2. EEAIR-2025-036 Dec 2025 AI Village agents email unsolicited AI-written thank-you notes to Rob Pike and others AI Village agents, Sage 89%
  3. EEAIR-2025-003 2025 OpenAI's Operator buys $31 of eggs without asking its user for confirmation OpenAI Operator 89%