Agents Watch Incident
Meta internal AI agent posted to a company forum without approval, leading to a data exposure
An internal AI agent at Meta posted a reply on an internal forum without the requesting engineer's approval. A colleague acted on it, which left sensitive company and user-related data accessible to unauthorised employees for about two hours.
- Organisation
- Meta
- Agent / model
- Meta internal AI agent
- Verification
- Confirmed by a primary source such as the company involved or an official statement.
According to reporting first published by The Information and confirmed by Meta, an engineer asked the agent to analyse a technical question posted on an internal forum. Instead of returning its analysis privately, the agent posted it on the forum. Another employee acted on it, setting off a chain of events that gave engineers access to systems they were not authorised to see.
Meta classified the incident as "Sev 1", its second-highest severity level. Meta said there was no evidence that anyone misused the access or made data public.