Skip to content

OpenAI

Prosecutors say chatbot prompts helped plan the OpenAI bug

5 min read

Written by an AI model from the linked sources and checked against them by a second model. How we work

All about: OpenAI models and products →
Image: latimes.com

Federal prosecutors have charged Abdulrahman Alfozan, a 35-year-old OpenAI engineer and Saudi national, with using or trying to use a device to intercept oral communications. The complaint, unsealed on October 13, says he hid a recorder in an executive conference room and points to his searches in a productivity tool, prompts from a chatbot account, office CCTV footage and a second recorder found in his bag, the Justice Department said in a press release.

Alfozan was arrested on October 9 and made his first appearance in federal court on Tuesday. If convicted, he faces up to 5 years in prison and a $250,000 fine, followed by 3 years of supervised release.

Behind the TV

OpenAI’s technical surveillance countermeasures team, the staff who sweep rooms for listening devices, found the recorder during a sweep on September 30. It was plugged into a USB port on the back of a television in an executive conference room in one of the company’s San Francisco buildings, wrapped in toilet paper. OpenAI concluded it had been installed on or about September 28, so it sat there for roughly two days before anyone found it. Its contents were encrypted.

The company removed the device for examination, closed the room, ran more sweeps and called the FBI. It also used its own models, working with forensic experts, on forensic images of the device’s storage. In its post on the incident, OpenAI said: “We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier.”

According to the complaint, meetings in that room often covered proprietary research, product plans and business-development opportunities, along with nonpublic information about OpenAI’s technical progress, strategic decisions, capabilities, government engagements and commercial relationships. Alfozan, who had worked at OpenAI as an engineer since July 2025, had no authority to enter the room or attend the meetings held there. Prosecutors say CCTV footage showed him near the room on multiple dates before the device was discovered.

The search trail

Before September 28, prosecutors say, Alfozan used a productivity tool to run several queries about OpenAI’s building addresses, meeting rooms, room capacities and calendars, “leadership and exec” locations and external partner meetings.

The chatbot evidence comes through a different account. It is associated with Alfozan’s spouse, but prosecutors say it shared multiple identifiers and attributes with his own chatbot account. Between September 24 and September 30, that account sent prompts about building, customizing, installing, testing, obfuscating and remotely accessing an audio recording device. The complaint quotes several of them, including “build[ing] a continuous audio recorder,” “customiz[ing] Wi-Fi . . . connectivity,” “improv[ing] audio and live listening” and “encrypt[ing] the SD card.”

On September 27 and 28, prosecutors say, the same account asked about Recording Device 1 itself, including its custom header and unique MAC address, the hardware identifier a network adapter uses on a local network. September 28 is also the day OpenAI says the device went in. And the prompts ran until September 30, the day the sweep found it.

A return visit and a second device

On October 5, five days after the recorder was pulled, Alfozan allegedly walked into the executive conference room and appeared to look behind the television where it had been plugged in. The complaint includes a CCTV still of the moment.

The same day, he left a bag unattended in a different room in an OpenAI building, prosecutors say. Security staff opened it and found his wallet and ID, among other things, plus a second recording device similar to the first. Four days later he was in custody.

OpenAI says it has fired Alfozan. “At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data,” the company said, adding that its security program is built to cover threats from “insider risks to cybercriminals to nation states.” The FBI investigation is ongoing.

How Washington framed it

Director of National Intelligence Jay Clayton said the case “underscores why winning the global super intelligence race is a national security prerogative,” and credited President Trump’s leadership in creating the “Super Intelligence Force.” FBI Director Kash Patel said the device was meant “to secretly capture sensitive communications about the company’s super intelligence technology” and called protecting American SI innovation one of the bureau’s “highest counterintelligence priorities.” U.S. Attorney Craig H. Missakian for the Northern District of California called it “a matter of national security.”

The Justice Department says the prosecution supports National Security Presidential Memorandum 11, or NSPM-11, which it describes as designed to support American superiority in super intelligence technology. Assistant U.S. Attorney Nicolas Parker is prosecuting, with help from Trial Attorney Kate Tagert of the National Security Division’s Counterintelligence and Export Control Section.

Alfozan’s attorney, William Pernik, said his client is presumed innocent and warned against seeking to “politicize” the prosecution based on factors such as nationality, per CBS News. He also cautioned against “jumping to conclusions at this very early stage of the case when very little information is known.”

Alfozan has been in custody since his arrest. His next court date is a detention hearing on October 19 before U.S. Magistrate Judge Alex G. Tse.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.