OpenAI engineer charged with planting hidden microphone in company's executive meeting room

A federal criminal complaint unsealed this week accuses an OpenAI engineer of secretly wiring one of the company's executive conference rooms for audio, in a case prosecutors describe as a national-security matter rather than routine workplace misconduct. OpenAI says its own security team discovered the device, used the company's AI models to help identify who planted it, and turned the case over to the FBI, which made an arrest within days.

The defendant, Abdulrahman Alfozan, 35, a Saudi national who joined OpenAI in July 2025 and worked on its AI agents platform, was arrested on Oct. 9 and made his first federal court appearance on Oct. 13 before Magistrate Judge Alex G. Tse in the Northern District of California, according to the unsealed federal complaint and the case's own court docket. He is charged with a single count of using an electronic device to intercept oral communications, a charge carrying a maximum of five years in prison, a $250,000 fine and three years of supervised release. He remains in custody.

A bug behind the television

The complaint says the device — a commercially available Seeed Studio microcontroller board that retails for about $14 — was plugged into a USB port behind a television in an executive conference room, wrapped in toilet paper to conceal it, and set to encrypt its recordings. OpenAI told TechCrunch the device had "advanced encryption, audio monitoring capabilities, and a mechanism for transferring data designed to evade monitoring."

According to the complaint, the installation did not happen in isolation. Between Sept. 24 and Sept. 30, an AI chatbot account tied to Alfozan's spouse — which investigators say shared identifying information with his own account — was used to ask questions about building a continuous audio recorder, encrypting its storage, and remotely accessing it, including the specific device's hardware identifier. Around the same period, prosecutors say Alfozan used an internal workplace tool to look up building addresses, meeting-room capacities, and the calendars and locations of leadership and partner meetings.

OpenAI's technical surveillance countermeasures team found the device on Sept. 30 during a routine sweep. The company began quietly restricting Alfozan's physical and network access that day while it investigated, and confirmed to NBC News that it placed him on leave and secured his equipment before his firing was finalized. On Oct. 5, the complaint alleges, he was recorded on a camera OpenAI had installed looking behind the same television. That same day, he allegedly left a bag unattended in another room; when security opened it, they found his wallet, his identification, and a second device matching the first.

The case by the numbers

• Sept. 24–30: alleged chatbot queries about building, hiding and remotely accessing a recorder
• Sept. 30: OpenAI's security sweep finds the first device
• Oct. 5: Alfozan allegedly caught on camera and a second device found in an unattended bag
• Oct. 9: fired and arrested by the FBI
• Oct. 13: first federal court appearance; detention hearing set for Oct. 19

The case is being prosecuted by the National Security, Cyber and Special Prosecutions Section of the U.S. Attorney's Office for the Northern District of California, under U.S. Attorney Craig Missakian. FBI Director Kash Patel has said the bureau treats protecting American AI firms from this kind of intrusion as among its highest counterintelligence priorities, and that the alleged scheme was aimed at capturing sensitive discussion of OpenAI's most advanced technology.

"For us it is a matter of national security," U.S. Attorney Craig Missakian said of the case.

A company increasingly on guard

OpenAI said in a statement posted to its official newsroom account that employees used its own AI models to analyze forensic images of the device and help identify the person who had planted it, before referring the matter to the FBI. The company wrote that it thanked investigators and prosecutors for moving quickly to make an arrest, and separately told reporters it has "no indications" that Alfozan accessed or removed model weights, source code or customer data. It said it does, however, have evidence he intended to place additional recording devices in other sensitive spaces, and that the conference room in question was used for confidential discussions of proprietary research, unreleased product plans and government engagements.

Before joining OpenAI, Alfozan worked as a founding engineer at the startup Nash from 2022 to 2025 and spent five years as a senior systems engineer at Meta, with earlier stints at Reach Labs, Microsoft, SAP and Akamai, and holds a master's degree in data science from UC Berkeley, according to the Standard's review of his professional history. Neither Alfozan nor a defense attorney has publicly commented on the charge; his family declined to speak with reporters outside their home in Millbrae, California.

The episode lands at a moment when OpenAI's security posture is under unusual scrutiny. The company has in recent months taken on sensitive government work, including agreements to run its models on classified networks, even as it races competitors to build what it calls "superintelligent" systems — the kind of stakes investigators say made its executive meetings an unusually attractive target. The case adds to a run of physical and cyber-security incidents the company has had to manage this year as its valuation and geopolitical profile have grown.

What happens next

Alfozan is due back before Judge Tse on Oct. 19 for a detention hearing, according to the docket, that will determine whether he remains in custody pending trial. No trial date has been set, and no plea has yet been entered on the public record. OpenAI has not said whether it plans any changes to its physical security procedures beyond the countermeasures sweeps that caught the device, though the case is likely to intensify pressure on AI labs more broadly to treat insider threats — not just external hacking — as a core part of protecting commercially and strategically sensitive work.