Federal agents have arrested an OpenAI worker after security staff uncovered a covert listening device concealed in an executive boardroom at the company's San Francisco headquarters.
Abdulrahman Alfozan, a Saudi citizen hired to the product team in July 2025, was dismissed and taken into custody on October 9. Federal prosecutors in the Northern District of California have charged him with intercepting oral communications, a felony that carries a maximum penalty of five years in prison.
Investigators allege the operation came unglued because of sloppy tradecraft. Staff found an encrypted audio recorder wrapped in tissue paper and plugged directly into a USB slot behind an office television around September 30. Prosecutors say Alfozan was not authorized to be in that room, which hosted confidential talks covering proprietary technical plans, commercial strategy, and government contracts.
OpenAI then checked security footage. Cameras captured Alfozan peering behind the television where the hardware was placed. Investigators also recovered an unattended bag on the company campus that held Alfozan's identification, his wallet, and a second recording device.
Chatbot Queries and Scoped Rooms
Prosecutors claim Alfozan systematically mapped the workplace before planting the bug around September 28. He reportedly tapped internal collaboration tools to pull office layouts, room capacities, executive calendars, and locations where top leadership gathered with outside partners.
Court filings also point to digital footprints left on a consumer chatbot account linked to his spouse, which shared identifiers with his own profile. Between September 24 and September 30, that account submitted queries asking how to build continuous voice recorders, obscure surveillance hardware, and encrypt storage cards.
OpenAI said on X that it deployed its own models to inspect the physical device and determine its origin. The company told reporters the gadget featured advanced encryption alongside data transfer components built to bypass monitoring systems. For customers, OpenAI maintained that the breach did not expose algorithmic weights, software codebases, or client records.
The arrest comes during heightened scrutiny across top frontier labs. Jacob Coxon, a former Anthropic staffer, recently claimed rival intelligence operations likely maintain assets inside both Anthropic and OpenAI. While authorities have not tied Alfozan's alleged actions to a specific foreign power, FBI director Kash Patel characterized the case as an effort to intercept superintelligence technology, calling the defense of domestic AI development a primary counterintelligence mandate.
Alfozan made an initial appearance in federal court in San Francisco. He has not entered a formal plea, and court records do not yet list defense counsel comments.
TBFiled by The Tech Business Desk
Big tech, startups, earnings, regulation and policy — the decisions that shape the products.
More from this desk →
Be the first to comment
Join the argument. No password, just your email or a passkey.