A Saudi national who worked at OpenAI as an engineer has been charged federally after a recording device was found hidden in an executive conference room at the company's San Francisco office, BBC reported. Abdulrahman Alfozan, aged 35, allegedly planted the encrypted device, which was wrapped in toilet paper and pushed into a USB port on the rear of a television in the room.
OpenAI employees found the device on 30 September. Prosecutors say it had been installed two days earlier. The company used its AI models to identify the device and trace it to a specific employee, then referred the matter to the FBI.
Alfozan was fired after the discovery and arrested on 9 October. The US Department of Justice published a statement, local time, detailing the complaint, and Alfozan appeared in court the same day.
FBI Director Kash Patel said Alfozan's alleged purpose was to capture information about OpenAI's superintelligence technology by hiding the device in the executive conference room. Alfozan had been working at OpenAI since July last year, according to prosecutors.
The conference room where the device was placed did not have a camera when it was installed, but CCTV footage elsewhere showed Alfozan near the room on about half a dozen occasions despite lacking authority to enter it. OpenAI later installed a camera in the room, which captured Alfozan entering and checking behind the television.
On 5 October, prosecutors say Alfozan returned to the conference room to inspect the device. That same day, he left an unattended bag elsewhere in the building. When a security team opened it, they found his wallet, identification and a second recording device similar to the first.
According to the FBI, Alfozan used a productivity tool to look up internal details including building addresses, meeting room locations, capacities and calendars, and leadership and executive locations. The Guardian reported that chatbot prompts attributed to Alfozan's spouse covered topics including building, installing and remotely accessing a recording device, customising Wi-Fi connectivity, improving audio and live listening, and encrypting an SD card. The same chatbot account made inquiries referencing unique identifiers of the device found in OpenAI's offices, The Guardian also reported.
OpenAI stated it had found nothing to suggest model weights, source code or customer data were improperly accessed or taken. The company expressed gratitude to agents and prosecutors for the speed of the arrest, and said the investigation continues. OpenAI said it takes protecting its technology, data and people seriously as a matter of national importance, and described its security program as designed to defend against a wide range of threats including insider risks, cybercriminals and nation states.
Prosecutors said the complaint did not set out what Alfozan intended to do with any recordings, and officials indicated they had not yet established his purpose. Whether Alfozan had retained legal representation was not immediately known, and he has not publicly commented on the indictment.
A conviction on the charge would expose Alfozan to a maximum of five years' imprisonment and a $250,000 fine. The case illustrates how AI tools can be used internally by companies to investigate security incidents, including identifying the source of a covertly placed device.
Reporting: BBC, ABC News, CNBC, The Guardian
Written by Pick & Scroll News from the reporting and documents linked above.
