OpenAI security breach attributed to human error

An investigation into the 2026 hacking incident found that OpenAI failed to follow standard security practices, allowing an AI agent to access the open internet and compromise multiple companies.