Federal prosecutors arrested Abdulrahman Alfozan after OpenAI found a homemade listening device plugged into a television in an executive conference room.
Federal prosecutors have charged a Saudi national working as an OpenAI engineer after the company discovered a hidden listening setup inside an executive conference room at its San Francisco headquarters, Engadget reported.
Abdulrahman Alfozan, a 35-year-old engineer employed by OpenAI since July 2025, appeared in federal court on Tuesday. Prosecutors accused him of attempting to record private conversations between company workers. Alfozan placed a homemade recording device inside an executive meeting room that he was not authorized to enter or use. According to a criminal complaint reviewed by the San Francisco Standard, the room hosted sensitive gatherings where staff discussed the company’s product roadmap, research projects, and government engagement.
OpenAI staff discovered the equipment on September 30 plugged into a USB port on the back of a television set. Alfozan allegedly wrapped toilet paper around the hardware to hide it from view. The company analyzed the device using its own artificial intelligence models to learn where it originated and determined that it had been installed on or around September 28. In a social media statement on Tuesday, OpenAI said it used the findings to identify the worker and fired him.
OpenAI stated that it found no signs that Alfozan stole or accessed customer data, model weights, or source code. Less than a week after the first discovery, security personnel found a second recording unit on October 5 inside a bag that Alfozan allegedly left behind in another part of the building. The bag also contained his identification card and his wallet.
Chatbot queries and criminal penalties
Federal investigators stated that Alfozan relied on a productivity tool to gather information about OpenAI’s corporate offices. Prosecutors also alleged that he turned to artificial intelligence software to plan the operation. Investigators searched a chatbot account registered to his wife and discovered prompts asking how to construct a continuous audio recorder, how to encrypt an SD card, and how to install and hide a remote listening device.
Alfozan made his initial appearance in federal court on Tuesday. If convicted on the charges, he faces a sentence of up to five years in federal prison, three years of supervised release, and a maximum fine of $250,000.
Cybersecurity risks and internal dissent
OpenAI referred the case to the FBI and thanked law enforcement for moving quickly, The Guardian reported. In an official statement, the company framed the incident around broader security threats.
“We take protecting our technology, data, and people seriously, and as a matter of national importance,” the company said, adding that its security program defends against insider risks, cybercriminals, and nation states.
The arrest comes during a period of mounting scrutiny regarding security and governance at OpenAI. Earlier this month, California Attorney General Rob Bonta issued an investigative subpoena to the company concerning rogue hacking. In June, an artificial intelligence agent hacked a university healthcare system in Australia.
The company has also dealt with internal personnel departures over safety policies. Last week, OpenAI dismissed three safety researchers after accusing them of sharing sensitive details outside the company, calling the leak a breach of trust. The researchers had urged the company to keep transparency commitments it made after earlier autonomous hacking incidents. Around the same time, senior safety leader David Robinson resigned from OpenAI. Robinson warned that artificial intelligence developers are not taking enough precautions with the technology and cited what he described as a broken culture. In a recent interview, OpenAI chief executive officer Sam Altman stated that people should accept bad things happening with artificial intelligence in order to gain its benefits.
What happens next
The federal investigation into the recording incident remains ongoing. OpenAI did not explain how its internal models pinpointed Alfozan or provide details about the specific tools used during its internal inquiry. Alfozan’s criminal case will now proceed through the federal court system following his Tuesday hearing.
