A security investigation at OpenAI has led to the arrest of an employee accused of secretly installing a listening device in a conference room used for confidential executive discussions. Federal prosecutors allege that the engineer attempted to capture sensitive conversations about the company’s artificial intelligence research and business operations.
Abdulrahman Alfozan, a 35-year-old Saudi national who joined OpenAI in July 2025, was arrested on October 9, 2026. The US Department of Justice announced the charges on October 13, accusing him of using or attempting to use an electronic device to unlawfully intercept oral communications.
The investigation began after OpenAI’s security team discovered a concealed recording device on September 30. Investigators determined that the equipment had been connected to a USB port behind a television in an executive conference room approximately two days earlier.
According to court documents, the small device was wrapped in toilet paper to conceal its presence and contained encrypted storage. The room was regularly used for sensitive meetings concerning proprietary research, upcoming products, technical developments, commercial partnerships, and government relationships.
Prosecutors allege that Alfozan gathered information about company facilities and executive schedules using an internal productivity tool. Investigators also identified activity on a chatbot account associated with his spouse, including questions about developing concealed audio recorders, wireless connectivity, remote monitoring, and encrypted storage.
Surveillance footage allegedly placed Alfozan near the restricted meeting room on several occasions despite his lack of authorization to attend meetings there. On October 5, cameras reportedly captured him returning to the room and examining the area behind the television.
That same day, security personnel discovered another recording device inside an unattended bag containing Alfozan’s identification and personal belongings. Investigators said the second device resembled the equipment previously recovered from the conference room.
OpenAI confirmed that its internal AI models assisted security personnel in analyzing the suspicious equipment and identifying the employee allegedly responsible. The company subsequently terminated his employment and referred the findings to the FBI.
Importantly, OpenAI said investigators had found no indications that the employee improperly accessed or extracted AI model weights, source code, or customer information. Authorities have also not publicly established his motive or confirmed what conversations, if any, were successfully recorded.
Alfozan faces a maximum prison sentence of five years and a $250,000 fine if convicted. His next court appearance is scheduled for October 19.