OpenAI Engineer Arrested Over Hidden Mic in Conference Room

Abdulrahman Alfozan, 35, an engineer on OpenAI's AI agents team, was arrested on Friday, October 9, prosecutors said Tuesday. OpenAI, the company behind ChatGPT, says he hid a recording device in a conference room used by its senior leadership. The device was wrapped in toilet paper and plugged into a USB port behind the TV. It turned out to be a Seeed XIAO ESP32-S3 Sense, a $24 microcontroller with a microphone, a camera, and an encrypted memory card. Between September 24 and 30, a chatbot account linked to his wife asked how to build a device for continuous recording, how to encrypt the SD card, and how to find that specific board's MAC address, a hardware identifier. OpenAI's security team found the device during a routine check of the conference room on September 30. The company fired Alfozan the day he was arrested, and he now faces up to five years in prison.

What's unsettling here is not the surveillance itself. It's that a chatbot helped put together a working bug out of questions that each sounded harmless on their own: card encryption, a board's MAC address, continuous recording. Assistants answer each question in isolation. They do not see the bigger picture behind them. Companies now need to watch not just what code an agent writes, but why an employee is asking about a specific piece of hardware. The device itself was found the old-fashioned way, during a routine physical check, not by AI. But once it turned up, OpenAI said it used its own AI models to figure out what the device was. Investigators used models too, to analyze forensic images of its memory. So the technology did end up doing real work later. It's just that plain old inspection caught the device, not a model.