The FBI has arrested an OpenAI engineer and charged him with planting a covert recording device inside an executive conference room at the artificial intelligence company’s San Francisco offices, in a case that has rattled the AI industry.
Abdulrahman Alfozan, 35, a Saudi national who had worked as an engineer at OpenAI since July 2025, was arrested on October 9, according to reports. He faces a federal charge of intercepting communications. OpenAI personnel found the thumb-sized device — fitted with a microphone and concealed in toilet paper — plugged into a USB port on the back of a television in the conference room on September 30; it had been installed around September 28, and its stored recordings were encrypted.
Caught by the company’s own AI
Investigators said surveillance footage showed Alfozan near the room several times before the device was found, though he had no authority to enter it, and that he returned on October 5, apparently to retrieve it. A search of his personal bag turned up a second, similar customised device. OpenAI said it used its own AI models to identify who had planted the recorder and trace where it had come from — and found evidence he intended to place more devices in sensitive areas of its offices.
The FBI director said, according to reports, that Alfozan had allegedly concealed the device to secretly capture sensitive communications about the company’s superintelligence technology. OpenAI said it had no indication that he had accessed or taken model weights, source code or customer data, and that the employee had been sacked.
AI labs are now espionage targets
The case is a sharp reminder that frontier AI labs have become targets for covert intelligence gathering, not just hacking. As companies and governments race for the talent and technology behind advanced AI, the meeting rooms where strategy and product plans are discussed are as valuable as the servers that train the models. An insider with a USB-port gadget can be harder to stop than a remote attacker — which is why, according to reports, OpenAI itself turned to its AI systems to hunt the culprit.
If convicted, Alfozan faces up to five years in prison and a $250,000 fine. The FBI investigation is continuing.
