OpenAI employee arrested over hidden recorder

Abdulrahman Alfozan, a 35-year-old OpenAI engineer, was arrested by the FBI on October 9 over a covert recorder found behind a television in an executive conference room at OpenAI’s San Francisco office, and made his first court appearance on October 13. The charge, using or attempting to use an electronic device to intercept oral communications, carries up to five years in prison and a $250,000 fine. What makes the case unusual is that AI sits on both ends: prosecutors say he planned the gadget with a chatbot, and OpenAI says its own models helped identify what the device was.

Key points

The recorder went into a USB port behind a conference-room TV around September 28, hidden inside toilet paper, and was caught by an OpenAI countersurveillance sweep on September 30.
A chatbot account linked to Alfozan’s spouse asked, between September 24 and 30, how to build a continuous audio recorder and encrypt an SD card, and asked about the device’s unique MAC address.
Wired identified the hardware as a Seeed XIAO ESP32-S3 Sense, a $13.99 consumer dev board; the claim that OpenAI’s internal AI models analyzed it comes only from OpenAI’s own statement.

Background

Alfozan joined OpenAI in July 2025 and, per his LinkedIn, led engineering on the AI agents platform. Before that he was a senior systems engineer at Meta from 2017 to 2022 and holds a master’s in data science from UC Berkeley. TechCrunch and Wired report he was fired the day he was arrested. The complaint adds an uncomfortable detail: he had used the company’s internal productivity tool to look up meetings scheduled in the targeted room, plus building addresses and the locations of “leadership and exec” personnel. Those queries are now part of the prosecution’s evidence.

The government has framed the case as counterintelligence rather than workplace spying. A trial attorney from the National Security Division’s Counterintelligence and Export Control Section is assisting the US Attorney’s office for the Northern District of California. FBI Director Kash Patel called protecting American “SI” innovation one of the bureau’s highest counterintelligence priorities, and the Justice Department ties the case to National Security Presidential Memorandum 11, aimed at supporting American superiority in super intelligence technology. US Attorney Craig Missakian put it plainly: “For us it is a matter of national security.” The “super intelligence” wording is the same shift we covered when federal documents started renaming AI (see our earlier piece on the super intelligence order).

Timeline

The complaint, unsealed October 13, lays out this sequence (all dates 2026):

Date	Event
Sept 24–30	A chatbot account tied to Alfozan’s spouse prompts for a continuous audio recorder, Wi-Fi customization, better live audio, and SD card encryption
Sept 27–28	The same account asks about Recording Device 1’s custom header and unique MAC address
~Sept 28	Device installed in the USB port behind an executive conference-room TV, concealed in toilet paper
Sept 30	Found during an OpenAI countersurveillance sweep; the FBI is notified
Oct 5	Surveillance footage catches Alfozan checking behind the TV; a second encrypted device turns up in the bag he left in a printer room
Oct 9	Arrested, and fired by OpenAI the same day
Oct 13	Complaint unsealed and initial appearance; Magistrate Judge Alex G. Tse holds the case, detention hearing set for Oct 19

The device and evidence

Wired identified the recorder as a Seeed XIAO ESP32-S3 Sense: a thumb-sized board with a detachable camera sensor, a digital microphone, and microSD storage, listed at $13.99 on Seeed’s site when Wired checked. OpenAI’s security team tried to export the recordings and failed; the contents were encrypted. The second device found on October 5 was another encrypted Seeed recorder, and the bag also held his wallet and ID. Wired kept one detail that reads like a farce beat: Alfozan later reported the bag missing and got it back, complained about a missing credit card, about $500 in cash, and two AirPods, and did not mention the second listening device.

What others say

OpenAI’s statement on X produced the version now circulating: employees “used the company’s AI models to determine what it was and where it came from,” and “at this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data.” OpenAI also called the device sophisticated, citing advanced encryption, audio monitoring, and a data-transfer mechanism designed to evade monitoring. TechCrunch attached its own caveat while relaying that: “it is unclear to what degree this device is technically distinct from other recording devices.” Wired’s account goes no further than the failed attempt to read the microSD card; no internal-model analysis appears anywhere in its reporting. Alfozan and his attorney had not responded to requests for comment. One more thing worth stating plainly: the complaint alleges no Saudi government connection. The counterintelligence framing is an official posture, not an established motive.

Our take

For any team rolling out AI productivity tools inside a company, this case teaches two lessons at once. The query logs of internal tools, meeting rooms, calendars, executive locations, are now part of the prosecution’s evidence, so the data boundaries for workplace agents should be drawn as if everything can be subpoenaed, because it can. And the physical side has escalated: a $13.99 dev board in a TV USB port anchored a Justice Department case at the highest counterintelligence priority. Take the “AI catches AI” narrative at a discount. It rests on OpenAI’s own account, while the evidence chain in the complaint is a countersurveillance sweep, CCTV, and a search of an unattended bag, which is classic physical security. The motive remains unknown. Less than two weeks after the chip-smuggling CEO arrest, the pattern repeats: a national-security frame arrives before the facts do. The next dated marker is the October 19 detention hearing.