OpenAI engineer charged with bugging exec conference room
US federal prosecutors have charged an OpenAI engineer, a Saudi national, with secretly placing an audio recording device in an executive conference room at one of the company's San Francisco buildings.
Abdulrahman Alfozan, 35, appeared in federal court on Tuesday. He faces one count of using or attempting to use a device to intercept oral communications. The charge carries a maximum penalty of five years in prison and a $250,000 fine.
According to prosecutors, the room regularly hosted meetings about sensitive proprietary research, product and business plans, government engagements and commercial relationships.
A recorder hidden behind a TV
The criminal complaint says Alfozan inserted "a surreptitious audio recording device" into a USB port at the back of a television in the conference room. The device was wrapped in toilet paper.
OpenAI staff found it on September 30, about two days after it had been installed. The recordings stored on it were encrypted.
The US Department of Justice released images of the television's USB port and of the device after it was removed.
Alfozan had joined OpenAI in July 2025. Prosecutors said security footage showed him near the room several times before the recorder was found. He had no authorization to enter the room or to attend the meetings held there.
Searches, calendars and chatbot prompts
Investigators say the groundwork started inside OpenAI's own systems. Before the device was installed, Alfozan allegedly used an internal productivity tool to look up building addresses, meeting-room calendars and the locations of "leadership and exec" meetings.
The complaint also points to a chatbot account linked to Alfozan's spouse. Prosecutors describe it as an account "which shared multiple identifiers and attributes with Alfozan's own chatbot account." That account submitted prompts that dealt specifically with building, installing and remotely accessing the recording device.
The complaint does not say which chatbot service was involved.
A second device in an unattended bag
The case did not end with the first discovery. About a week after the recorder was found, security cameras recorded Alfozan entering the conference room and apparently checking behind the television.
On the same day, OpenAI security found a bag he had left unattended. It held his wallet, his ID and a second recording device similar to the first one.
FBI Director Kash Patel commented on the case, saying that "protecting American SI [super intelligence] innovation is essential to safeguarding our nation's economic strength and national security." "Super intelligence" is the term President Donald Trump prefers to use for artificial intelligence.
No foreign government named
Despite Alfozan's nationality and the wording of Patel's statement, prosecutors have not accused him of working for a foreign government. They also have not identified who, if anyone, was meant to receive the recordings.
That leaves the motive open. The charge itself concerns the interception of oral communications, not espionage or theft of trade secrets.
Our Take
For security teams, the most telling detail is how low-tech the alleged operation was. There was no zero-day and no phishing campaign, just a small recorder plugged into a TV and wrapped in toilet paper. AI labs spend heavily on securing model weights and cloud systems, but this case suggests that meeting rooms, AV equipment and the USB ports on them deserve a place in threat models too.
The case also fits a wider pattern of insider risk. Staff with legitimate access remain hard to detect, as earlier cases like the engineer jailed for locking 3,000+ devices and reports of agents sent into US law firms have shown. It is worth noting that the alleged reconnaissance relied on ordinary internal tools such as calendars and directory searches. Monitoring for unusual lookups of executive schedules could be a practical lesson for other companies.
The chatbot angle is notable as well. Prompt logs appear to have become evidence, which mirrors how OpenAI has used account activity to ban accounts tied to state operations. AI usage records may increasingly turn up in criminal investigations.
It is worth watching whether prosecutors later name a recipient for the recordings or file additional charges. Until then, the national security framing used by the FBI goes further than the allegations actually made in court.