A concealed device was allegedly installed on or around September 28 in an executive conference room at OpenAI’s San Francisco office. It was plugged into a television’s USB port and reportedly covered in toilet paper — an unusually crude disguise for equipment authorities describe as part of a serious alleged breach.¹

The room was not incidental. According to the Justice Department’s account of the complaint, meetings there often covered proprietary research, product plans, commercial opportunities and nonpublic information about OpenAI’s technical and strategic decisions.²

OpenAI says it responded by turning to its own technology. The company said it used its models to “understand the device and where it had come from,” concluding that a specific employee had planted it days earlier.¹ That claim puts a striking twist on a conventional insider-security case: an AI company presenting AI not as the object of a breach, but as an investigative tool.

Federal authorities then arrested Abdulrahman Alfozan, 35, a Saudi national who had worked at OpenAI since July 2025, on October 9. He was charged with using an electronic device to intercept oral communications and made his first federal court appearance the following Tuesday.³

OpenAI fired Alfozan and referred the matter to the FBI, saying the investigation remains active. The company stressed that it had “no indications” he accessed or exfiltrated model weights, source code or customer data.¹ Its broader message is that the episode demonstrates how AI can counter “sophisticated security threats”; the criminal case, however, will hinge on the more traditional question of whether the device was used to secretly capture private conversations.²