OpenAI Engineer Arrested for Bugging Executive Conference Room

A 35-year-old Saudi national employed as an engineer at OpenAI was arrested Oct. 9 after federal authorities accused him of planting a covert audio recording device inside the artificial intelligence company’s San Francisco headquarters.

Abdulrahman Alfozan faces a single federal charge of using or attempting to use a device to intercept oral communications, according to the Department of Justice. The charge carries a maximum penalty of five years in prison, a $250,000 fine and three years of supervised release upon conviction. Alfozan made his initial appearance in federal court Tuesday.

The targeted space was an executive conference room restricted to senior leadership and used exclusively for highly confidential discussions regarding proprietary research, product roadmaps and business opportunities. OpenAI is currently developing what it refers to internally as superintelligence technology, a pursuit that has drawn intense scrutiny from U.S. counterintelligence agencies concerned about foreign espionage targeting American AI laboratories.

Security personnel discovered the encrypted recording device Sept. 30 plugged into a USB port on the back of a television monitor inside the restricted room. Prosecutors allege the hardware had been installed two days prior and concealed within toilet paper.

Internal surveillance footage showed Alfozan lingering near the executive conference room multiple times before the discovery, despite having no authorization to enter the space. Investigators also found he had conducted extensive internal searches mapping out OpenAI’s buildings, meeting rooms, executive calendars and external partner schedules.

A second, similar recording device was recovered Oct. 5 inside a bag Alfozan left unattended in another room alongside his wallet and identification, the DOJ said. Surveillance footage from that same day captured him entering the executive conference room and appearing to inspect the area behind the television where the first device remained hidden.

Digital forensics linked the alleged operation to a chatbot account associated with Alfozan’s spouse. Authorities say the account shared identifiers with Alfozan’s own profile and submitted detailed prompts asking how to build and conceal continuous audio recorders equipped with Wi-Fi connectivity, live listening capabilities and encrypted storage. The technical specifications requested in those prompts matched the devices ultimately recovered by OpenAI security.

Alfozan had worked at the company since July 2025. Following the discovery, OpenAI terminated his employment and referred the case to the FBI.

“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 company said in a statement. OpenAI confirmed its investigation found no evidence that Alfozan accessed or exfiltrated model weights, source code or customer data.

FBI Director Kash Patel characterized the arrest as part of a broader mandate, stating on X that protecting American superintelligence innovation ranks among the bureau’s highest counterintelligence priorities.

The incident underscores a persistent vulnerability for frontier AI developers operating in highly competitive, globally contested markets. Unlike traditional software companies, firms like OpenAI, Anthropic and DeepMind hold intellectual property tied directly to national security interests. Corporate espionage in this sector frequently involves insider threats rather than external network breaches, making physical access controls and personnel vetting critical layers of defense.

This arrest follows separate recent turmoil over information security at the company. OpenAI recently fired three safety researchers accused of sharing confidential internal information, signaling ongoing friction between the firm’s rapid expansion and its ability to maintain strict operational secrecy.

Amy Kremer, chairwoman of Humans First, argued the allegations expose systemic gaps in how advanced AI labs secure their facilities. “If these companies want Americans to believe that winning the AI race is a national security imperative, securing their own labs must be a basic requirement,” Kremer told Fox Business, calling for enforceable security standards, meaningful background screening and strict controls on access to sensitive research.

Alfozan remains in federal custody pending further proceedings.