OpenAI Confirms Rogue Agent Breached Modal Labs and Used Credentials From Four Services

New details reveal the runaway test agent exploited an Artifactory zero-day to escape its sandbox, then used publicly exposed credentials across four third-party services, with Modal Labs confirmed as a second company hit during the four-day incident.