OpenAI employee accused of planting hidden recorder in room used for confidential meetings

• OpenAI engineer Abdulrahman Alfozan was arrested on October 9 and charged over an alleged recording device.
• The device was reportedly concealed in toilet paper and inserted into a USB port behind a television.
• OpenAI says it found no evidence that model weights, source code or customer data were compromised.

OpenAI has used its own artificial intelligence (AI) tools to help investigate a hidden recording device discovered inside its San Francisco office, in a case that has led to the arrest of one of its engineers.

Abdulrahman Alfozan, a 35-year-old Saudi national employed by OpenAI since July 2025, was arrested by the FBI on October 9. He is accused of using or attempting to use an electronic device to intercept oral communications, according to the US Department of Justice (DOJ).

The device was discovered on September 30 in an executive conference room, where meetings frequently involved highly confidential technical, commercial and strategic information. OpenAI determined that it had been installed around September 28 and that its contents were encrypted.

The company reported the incident to federal authorities and used its AI capabilities to help analyse the equipment and investigate its origins. Alfozan made his initial appearance in federal court on October 13. The allegations have not been proven, and he is presumed innocent unless convicted.

How was the recording device discovered?

According to the DOJ, the device had been inserted into a USB port on the back of a television in the executive meeting room. It was concealed in toilet paper, making it less conspicuous.

Security footage reportedly showed Alfozan near the room on multiple occasions before the device was discovered, despite him having no authority to access or participate in meetings held there.

Investigators also examined inquiries allegedly made by Alfozan using a productivity tool about OpenAI’s building addresses, meeting rooms, room capacities, calendars and the locations of senior executives and external partner meetings.

The complaint further alleges that a chatbot account associated with his spouse was used between September 24 and September 30 to ask questions about building, installing, concealing and remotely accessing an audio recording device. Some prompts reportedly concerned continuous audio recording, Wi-Fi connectivity, live listening and encrypting data stored on an SD card.

Investigators allege that the same account submitted prompts linked to the specific device found at OpenAI. These details form part of the allegations in the criminal complaint and have not been established as proof of guilt.

A second recording device was also found in a bag left unattended in another room at an OpenAI building, according to the DOJ.

What information could have been exposed?

The executive conference room was regularly used to discuss sensitive information, including proprietary research, product plans, business development, technical progress, strategic decisions, government engagements and commercial relationships.

OpenAI has said it found no evidence that Alfozan had successfully obtained or misused its model weights, source code or customer data. The company also confirmed that the employee had been terminated and said it would continue co-operating with federal investigators.

The case highlights a potential security risk for technology companies: sensitive information can be discussed in physical meeting rooms, leaving it vulnerable to covert recording even when digital systems remain secure.

If convicted, Alfozan faces a maximum penalty of five years in prison and a $250,000 fine, followed by a period of supervised release. The FBI investigation is continuing.