Washington presented Tuesday as a national security matter the extremely rare arrest of a Saudi engineer from OpenAI, accused of having hidden a microphone in a management meeting room, prosecuted at this stage for “interception of communications” and not for espionage.
Abdulrahman Alfozan, 35, employed since July 2025 by the creator of ChatGPT, was arrested on October 9 and appeared Tuesday before a federal judge in San Francisco for “interception of communications,” a misdemeanor punishable by up to five years in prison, the Justice Department announced. He remains detained pending a hearing on October 19.
“This serious case shows why winning the global race for superintelligence is a national security imperative,” said in a statement the director of American intelligence, Jay Clayton, placed at the beginning of October at the head of the “Super Intelligence Force”, a working group on AI from the White House, which is trying to impose the renaming of this technology “super intelligence”.
“For us, this is a question of national security,” added federal prosecutor Craig Missakian. Jay Clayton spoke of the fight against “foreign threats”, without naming a sponsor in this affair. At this stage, the ministry does not accuse the engineer of economic espionage or theft of industrial secrets.
“We have no indication that this person inappropriately accessed model weights, source code or customer data,” an OpenAI spokesperson told AFP.
Weights are the parameters that determine how a model works.
Hidden in toilet paper
According to the complaint, summarized by the ministry, the device was discovered on September 30 by OpenAI security teams, plugged into the USB port of a television and hidden in toilet paper. Installed two days earlier, it contained encrypted records.
The meetings held in this room focused in particular on research, product projects and discussions with public authorities.
Before installing it, the engineer had inquired via a productivity tool about the meeting rooms, their agendas and the meeting places of management, claims the accusation.
Between September 24 and September 30, an AI assistant account linked to his wife sent requests about making an audio recorder, listening to live streams or encrypting a memory card, some of which matched the discovered item, according to the complaint. A second similar device was discovered in a bag belonging to him.
An elder from Meta
OpenAI, which reported the matter to the FBI, claims to have used its own AI models to analyze the device and trace it back to the employee. The company says it gradually restricted his access before firing him on October 9, in coordination with investigators.
According to a LinkedIn profile in his name, Abdulrahman Alfozan, a graduate of the Massachusetts Institute of Technology (MIT) and holder of a master’s degree from Berkeley (California), worked for five years at the social media giant Meta then for a logistics start-up before becoming head of engineering on OpenAI’s AI agent platform. In a short video posted in September on YouTube, he was presented as the company’s only Saudi employee.
The case comes as the Trump administration has made American supremacy in AI a national security priority, including through a presidential memorandum signed in June, which the Justice Department cites in support of the prosecution.
In January, a former Google engineer, Linwei Ding, was convicted of stealing AI-related company secrets while working concurrently for two Chinese companies. He was charged in 2024 with theft of industrial secrets, a charge punishable by ten years in prison.
