An OpenAI engineer faces a federal charge of intercepting oral communications over an audio recording device allegedly hidden in an executive conference room at the company’s San Francisco offices. The complaint says the device, found Sept. 30, was plugged into a USB port behind a television and concealed in toilet paper. The complaint alleges the room hosted confidential discussions about proprietary research and government engagements and that security personnel later found a second, similar recorder in a bag the engineer left unattended, together with the engineer’s wallet and ID.

Key Takeaways

AI-generated summary, reviewed by an editor. More on our AI guidelines.

The device and the room

The Justice Department identifies the engineer as a 35-year-old national of the Kingdom of Saudi Arabia, arrested Friday, Oct. 9. The complaint unsealed Oct. 13 says OpenAI determined the device was installed around Sept. 28. Prosecutors allege surveillance footage showed the engineer near the room before its discovery, despite lacking authorization to enter or attend meetings there.

Craig Missakian, the U.S. attorney for the Northern District of California, said: “For us it is a matter of national security.”

FBI Director Kash Patel said the complaint alleges the engineer installed the device “to secretly capture sensitive communications about the company’s super intelligence technology.”

Inquiries and a second device

Before Sept. 28, the engineer used a productivity tool to ask about building addresses, room capacities and calendars, including leadership and external partner meetings, the complaint alleges.

Prosecutors say a chatbot account associated with the engineer’s spouse shared identifiers and attributes with the engineer’s account. Between Sept. 24 and Sept. 30, the day the device was found, it submitted prompts about building, concealing and remotely accessing a recorder, including “build[ing] a continuous audio recorder” and “encrypt[ing] the SD card.”

On Sept. 27 and Sept. 28, the account submitted prompts about the specific device later found in the conference room, including its custom header and unique MAC address, prosecutors say.

The Justice Department’s releases do not name the chatbot.

The complaint alleges the engineer entered the room Oct. 5 and appeared to look behind the television. On or about Oct. 5, the engineer allegedly left a bag unattended in another room; security personnel who opened it found the engineer’s wallet and ID and a second, similar recording device, prosecutors say.

OpenAI’s investigation

OpenAI said its surveillance countermeasures team discovered the recorder during a Sept. 30 sweep. The company removed it, closed the room, conducted further sweeps and contacted the FBI.

The company said it used its models alongside forensic experts to examine images of the device’s storage and assess changes in the data.

OpenAI said it restricted the engineer’s access in stages after the discovery, coordinated the firing with the FBI, and terminated the engineer Oct. 9, the day of the arrest.

OpenAI said it has no indications of improper access to or exfiltration of model weights, source code or customer data.

Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.

What the record does not establish

OpenAI calls the device “sophisticated,” with “advanced encryption, audio monitoring capabilities, and a mechanism for transferring data designed to evade monitoring.”

The complaint describes a recorder plugged into a television and wrapped in toilet paper. A complaint photo shows a board bearing Seeed Studio branding. The alleged chatbot prompts asked how to build a recorder. It is unclear how the device differs technically from other recording devices.

The charge is interception of oral communications, not espionage or trade-secret theft. The public record does not identify any intended recipient. The device’s contents were encrypted; nothing public establishes what it captured.

The engineer and an attorney for the engineer could not be reached for comment. A complaint merely alleges crimes, and the defendant is presumed innocent until proven guilty beyond a reasonable doubt.

Detention hearing

The engineer appeared in federal court Oct. 13 and remains in federal custody. If convicted, the engineer faces up to five years in prison and a $250,000 fine, followed by three years of supervised release.

A detention hearing is scheduled for Oct. 19 before U.S. Magistrate Judge Alex G. Tse.

Frequently Asked Questions

What is the OpenAI engineer charged with?

Interception of oral communications, according to a federal complaint unsealed Oct. 13. The charge is not espionage or trade-secret theft. If convicted, the engineer faces up to five years in prison and a $250,000 fine, followed by three years of supervised release.

Where was the recording device found?

In an executive conference room at OpenAI's San Francisco offices. The complaint says the device was plugged into a USB port behind a television and concealed in toilet paper. OpenAI said its surveillance countermeasures team discovered it during a Sept. 30 sweep and determined it was installed around Sept. 28.

What evidence does the complaint describe?

Surveillance footage of the engineer near the room, productivity-tool inquiries about buildings, rooms and calendars, and chatbot prompts from an account associated with the engineer's spouse about building a recorder. It also alleges a second, similar device was found in a bag the engineer left unattended, alongside the engineer's wallet and ID.

Did the device capture anything or was OpenAI data taken?

Nothing public establishes what the device captured, because its contents were encrypted. The public record does not identify any intended recipient. OpenAI said it has no indications of improper access to or exfiltration of model weights, source code or customer data.

What happens next in the case?

The engineer appeared in federal court Oct. 13 and remains in federal custody. A detention hearing is scheduled for Oct. 19 before U.S. Magistrate Judge Alex G. Tse. A complaint merely alleges crimes, and the defendant is presumed innocent until proven guilty beyond a reasonable doubt.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Mistral Previews 1-Trillion-Parameter Large 4, Pitching Open Weights Against Lock-In
Earlier this year, Hugging Face was breached by rogue OpenAI agents. It turned to leading American closed models to help defend itself, only to find their guardrails blocking the work. The guardrails
The Implicator
Anthropic Details Claude’s Role in Surveillance in Mali, China and Iran
Anthropic said Thursday that it had blocked Claude accounts tied to state-aligned actors and contractors who used its models to help build or run surveillance operations in Mali, China and Iran betwee
The Implicator
OpenAI Models Ran a Hack in Hours That Takes Skilled Humans Weeks
OpenAI's advanced models breached Hugging Face's internal systems in hours, an attack that would typically take a skilled human a couple of weeks. People familiar with the matter gave that account to
The Implicator
AI News

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai