OpenAI Engineer Charged Over Recorder Wrapped in Toilet Paper

Federal prosecutors have charged OpenAI engineer Abdulrahman Alfozan, 35, with using or trying to use a device to intercept oral communications, after a recorder was found in an executive conference room. ^[2]

The story is developing and rests on reporting, not on court papers the paper has seen. The San Francisco Standard, citing prosecutors and a criminal complaint, reported the arrest on October 13, 2026, and The Verge followed with OpenAI's public statement. ^{[1] [2]} A charge is an accusation.

He made an initial federal court appearance on October 13 and remained in custody, according to the Standard. ^[2]

Where the device was found

OpenAI personnel found the device on September 30, the Standard reported. It was plugged into a USB port on the back of a television in an executive conference room and wrapped in toilet paper. The company worked out that it had been installed around September 28, and its contents were encrypted. ^[2]

The complaint, as the Standard describes it, says meetings in that room often covered confidential technical, commercial and strategic matters, including proprietary research, product plans and government engagements. It says surveillance footage showed Alfozan near the room on several dates, and that he was not authorized to enter it or attend meetings there. ^[2]

A second piece of evidence turned up on October 5, according to the complaint. That day Alfozan allegedly entered the conference room and appeared to look behind the television. Security staff allegedly found an unattended bag in another room, holding his wallet and identification and a second, similar recording device. ^[2]

The prompts

The most striking part of the complaint, as the Standard reports it, concerns what was typed into software. It alleges that Alfozan used a productivity tool to ask about OpenAI buildings, meeting rooms, room capacities, calendars, and where leaders and outside partners would be meeting. ^[2]

It also alleges that a chatbot account tied to his spouse, sharing identifiers with his own, was used between September 24 and September 30 to ask how to build, install, hide and remotely reach a recording device. The prompts allegedly included how to "build a continuous audio recorder" and how to "encrypt the SD card." ^[2]

These are allegations in a complaint, relayed secondhand, not findings.

What OpenAI says

OpenAI told The Verge it found a device, used its own AI models to help understand it and trace where it came from, referred the matter to the FBI, cooperated with investigators and fired Alfozan. ^[1] The Standard added that the company coordinated the timing of the firing with the FBI, and first restricted his access gradually before suspending his physical and network access. ^[2]

The company described the device as sophisticated and built to evade detection, with encryption, audio-monitoring capability and a way to move data off it that was meant to avoid monitoring. Those are OpenAI's words, and no outside examination of the device has been reported. ^[2]

Then comes the part the company most wants readers to hold onto. OpenAI said it has no indication that Alfozan improperly accessed or took model weights, source code or customer data, and has seen no activity compromising product safety or customer information. ^{[1] [2]} Model weights are the billions of stored numbers that make up a trained AI system, the closest thing a lab has to a crown jewel.

That reassurance is narrower than it sounds. It speaks to what happened on OpenAI's servers. A microphone in a meeting room is a different kind of leak, and it does not touch any of those systems. Nothing in the reports says what, if anything, the device recorded.

Why prosecutors are loud about it

U.S. Attorney Craig Missakian called the case a matter of national security, the Standard reported. FBI Director Kash Patel said protecting American AI innovation was among the bureau's highest counterintelligence priorities, meaning its work against spying and covert collection of secrets. ^[2]

Their statements show that officials are treating this case as a national security matter. The complaint says the targeted room was used for discussions of government engagements. ^[2] But the officials' words describe how seriously they take the case, not what the evidence shows about motive. The reports do not say why the device was allegedly placed, or for whom.

There is an irony the company will not enjoy. OpenAI builds systems that summarize meetings, search calendars and answer questions about the world around them. Its security team's own tools were reportedly used to understand a gadget that, if the complaint is right, sat behind a television wrapped in toilet paper. The most guarded building in AI was apparently exposed through a spare USB port.

What is not known

The reports do not establish that the device captured a meeting, that its contents were retrieved or sent anywhere, or what the second device was meant to do. They do not say whether either device ever connected to a network. Alfozan's defense has not been described, and the complaint itself has not been made available for checking. ^{[1] [2]}

The FBI investigation is ongoing, according to OpenAI. ^[1] The next step in court is a detention hearing before U.S. Magistrate Judge Alex G. Tse, scheduled for October 19. ^[2]