A Saudi national working as an OpenAI engineer was arrested by the FBI after planting an encrypted recording device hidden inside a roll of toilet paper in the company's executive conference room — and the company that wants to decide what you're allowed to say online needed its own AI just to figure out who was spying on it.
Abdulrahman Alfozan, 35, was arrested October 9 and charged with installing the concealed device, which he plugged into a TV in a San Francisco conference room where confidential meetings regularly take place, according to the Justice Department. The device was discovered during a security sweep two days after it was planted on September 28. Surveillance footage allegedly captured Alfozan casing the room in the days before the device was found.
Here's the wrinkle that should bother every American: OpenAI used its own AI models to trace the device and identify Alfozan. The company that can't secure its own executive suites from a guy hiding a bug in a bathroom roll is the same company building tools it insists should shape the boundaries of acceptable speech online.
The indictment gets worse. Alfozan allegedly prompted AI chatbots for internal OpenAI details — building addresses, meeting room capacities, executive locations and calendars. His spouse, according to the FBI, used the same chatbot account to ask about building continuous audio recorders, customizing WiFi connectivity, improving live listening capability, and encrypting SD cards. Some of those queries referenced unique identifiers matching the device found in the office. When FBI agents searched a bag Alfozan left unattended at headquarters, they found a second identical recording device.
OpenAI says there's no evidence Alfozan accessed model weights, source code, or customer data. He was fired. If convicted, he faces up to five years in prison and $250,000 in fines. He has not publicly commented.
The Guardian noted the incident adds to OpenAI's growing list of privacy controversies, including a June incident where an AI agent hacked Australia's university healthcare system and a recent investigative subpoena from California's attorney general over autonomous hacking. HuffPost reported that OpenAI told investors its annualized revenue for September was nearly $50 billion — a drop from earlier signals.
Then there's the timing. OpenAI just fired three safety researchers — Jasmine Wang, Tomek Korbak, and Mikita Balesni — for what the company called "a breach of trust" over handling sensitive information. The researchers published a letter arguing the abrupt dismissals undermine a culture of raising safety concerns. So the company that can't catch a physical bug in its own conference room without AI assistance is also silencing the people tasked with flagging its risks.
A Saudi national. An encrypted device. Executive meeting rooms. A second device waiting in a bag. The FBI is still investigating. The question isn't just what Alfozan was recording — it's who he was recording it for, and why a company this porous thinks it should be trusted to police anyone's speech.
