OpenAI says a hidden recording device discovered in an executive conference room at its San Francisco office has led to the arrest and dismissal of a company engineer — a case the firm is presenting as an unusually stark example of AI being turned inward to protect itself.

The device was allegedly installed on or around September 28, plugged into a television’s USB port and concealed with toilet paper, according to reporting on the federal complaint. CCTV footage reportedly placed Abdulrahman Alfozan near the room on several dates before the device was found.¹

OpenAI said it then examined the equipment with its own models. “We used our models to understand the device and where it had come from,” the company said, concluding that “a specific employee had planted it a few days earlier.”² The company referred the matter to the FBI and fired the employee.

The Justice Department identified Alfozan, 35, as a Saudi national charged with using an electronic device to intercept oral communications. He was arrested on October 9 and made his first federal court appearance Tuesday, according to the report. The meetings held in the room frequently involved proprietary research, product plans, commercial opportunities and nonpublic discussions of OpenAI’s technical progress and strategic decisions.¹

OpenAI’s account stresses the limits of the apparent breach as well as its seriousness. The company said it had found no indication that Alfozan accessed or removed model weights, source code or customer data. Yet it framed the alleged device as part of a broader security challenge spanning “insider risks to cybercriminals to nation states.”²

The FBI investigation remains ongoing. For OpenAI, the episode lands amid wider scrutiny of internal controls after separate, unrelated dismissals of safety and alignment staff — making the alleged bugging case not just a criminal inquiry, but another test of trust inside one of AI’s most closely watched companies.¹