Employees discovered the device on Sept. 30 inserted into a USB port behind a television. It had reportedly been concealed in toilet paper and installed around Sept. The room was used for meetings involving confidential research, product plans, business opportunities, company capabilities, government engagements and commercial relationships.
The criminal complaint alleges that Alfozan had appeared near the room on multiple occasions despite lacking authorization to enter it or attend its meetings. After OpenAI added a camera, footage from Oct. 5 allegedly showed him entering the room and looking behind the television. Security personnel also found a second, similar recording device in an unattended bag that contained Alfozan’s wallet and identification.
Investigators further allege that Alfozan used workplace tools to examine office addresses, meeting rooms, capacities, calendars and locations associated with executives and outside partners. A chatbot account linked to his spouse allegedly submitted prompts about building, concealing, testing, encrypting and remotely accessing an audio recorder between Sept. 24 and Sept. 30, including prompts tied to the device’s identifying details.
OpenAI said it used its models to analyze the device and identify the employee it believed had planted it, then referred the matter to the FBI and terminated him. The company said it had no indication that Alfozan accessed or removed model weights, source code or customer data. The FBI investigation remains ongoing.
Federal prosecutors described the alleged conduct as a national-security concern because the room contained sensitive information about advanced artificial-intelligence work. Alfozan is presumed innocent. If convicted under the charge, he could face up to five years in prison, a $250,000 fine and three years of supervised release.