An OpenAI engineer is in federal custody after allegedly hiding an encrypted audio recorder behind a television in one of the company's executive conference rooms in San Francisco. OpenAI says its own AI models helped figure out who put it there.
“We discovered a sophisticated recording device that was hidden inside one of our offices,” OpenAI said in a statement on X. “We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier.”
The Justice Department identified the suspect as Abdulrahman Alfozan, a 35-year-old Saudi national who has worked as an engineer at OpenAI since July 2025. He was arrested on Oct. 9, charged with using or attempting to use a device to intercept oral communications, and made his initial appearance in federal court on Tuesday.
“We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest,” OpenAI said. “The FBI’s investigation is ongoing and we will continue to support their work.”
According to the criminal complaint, OpenAI staff found the recorder on or around Sept. 30, plugged into a USB port on the back of a TV and wrapped in toilet paper. Its contents were encrypted, and the company determined it had been installed around Sept. 28.
Security cameras allegedly caught Alfozan near the room on several dates before the discovery, even though he had no authorization to enter it or attend its meetings. Prosecutors say those meetings regularly involved confidential technical and strategic information.
Before the device went in, Alfozan allegedly used an internal productivity tool to ask about OpenAI's building addresses, meeting rooms, room capacities, calendars, and where leadership and outside partners met.
Prosecutors also point to a chatbot account tied to Alfozan's spouse, which shared multiple identifiers with his own account. Between Sept. 24 and 30, it allegedly asked how to build a continuous audio recorder, add Wi-Fi connectivity, improve live listening, and encrypt the device's SD card. The complaint doesn't say whose chatbot it was.
“At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data,” the company said, referring to the trained parameters that make up an AI model, its underlying code, and user information. “The employee was terminated.”
“We take protecting our technology, data, and people seriously, and as a matter of national importance,” OpenAI added, saying its security program is built to defend against threats ranging from insiders to cybercriminals to nation states. “This case is an example of how AI can help address these sophisticated security threats.”
The company did not explain in its statement exactly what its models analyzed to identify him.
Federal officials are treating it as a national security case. FBI Director Kash Patel said Alfozan allegedly tried to capture sensitive conversations about the company's “super intelligence technology,” and U.S. Attorney Craig Missakian called the prosecution a national security matter.
The arrest came days after OpenAI's newsroom account defended firing three safety researchers—Jasmine Wang, Tomek Korbak, and Mikita Balesni—saying an investigation found they had violated clear policies on handling sensitive information.
OpenAI's security record has drawn scrutiny since July, when the company admitted its own models escaped a test environment and breached Hugging Face's infrastructure to grab benchmark answers.
If convicted, Alfozan faces up to five years in prison, a $250,000 fine, and three years of supervised release. A detention hearing is set for Oct. 19 before U.S. Magistrate Judge Alex G. Tse.
