What the complaint alleges
Federal prosecutors have charged Abdulrahman Alfozan, a 35-year-old Saudi national who joined OpenAI as an engineer in July 2025, with attempting to intercept oral communications. He was arrested on 9 October and made his first appearance in federal court in San Francisco on 13 October. The offence carries up to five years in prison and a $250,000 fine. Everything that follows comes from a criminal complaint and has not been proven in court.
The Justice Department says a recording device was installed in an executive conference room at an OpenAI building in San Francisco on or about 28 September and found on or about 30 September, wrapped in toilet paper and connected through a USB port on the back of a television. The room is used for non-public technical meetings and for meetings with government representatives. Officials have not said whether the device captured anything.
OpenAI analysed the device with its own models. According to the complaint it may have been modified to block data moving between the recorder and the television, and its microSD card may be encrypted: the company made a forensic image and could not retrieve any data from it.
A second device, and a bag reported missing
On 5 October, prosecutors say, Alfozan left a bag unattended in an office printer room. Security staff opened it and found a wallet, an ID card and a second recording device of the same type. He then reported the bag missing, collected it from lost and found the next day, and filed a written report listing items he said had gone: a credit card, about $500 in cash, two pairs of AirPods, headphones, sunglasses and a power bank.
The conference room had no camera covering it when the first device was planted. OpenAI installed one afterwards, and prosecutors say an image from 5 October, taken by a camera behind the television, shows Alfozan checking on the device. Office CCTV places him near the room on several dates, although he had no authority to enter it or attend the meetings held there.
AI tooling on both sides of the case
The complaint describes two separate uses of AI tools by the defendant. Between roughly 17 August and 16 September, prosecutors allege, he used a productivity agent to ask about company building addresses, meeting rooms, room capacities, calendars and the venues of meetings with external partners. Then, between 24 and 30 September, a chatbot account linked to his spouse allegedly received prompts about building and modifying an audio recording device, covering recording reliability, browser access and controls, and a second unit that would start recording on speech. That account shared installation identifiers, IP usage logs and user-path references with his own, prosecutors say.
On the other side of the investigation, OpenAI said it used its internal models to identify the person responsible and trace the device’s origins. The company said it took gradual steps to contain the risk while the investigation continued, restricting his access and placing him on leave before terminating his employment on 9 October, and that it referred the matter to the FBI. It said it has evidence the employee intended to deploy other recording devices, and added: “At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data.”
What to watch
FBI Director Kash Patel said protecting American work in the field “is essential to safeguarding our nation’s economic strength”. The bureau says its investigation is continuing. Alfozan has been a US permanent resident since September 2020 and is due back in court on 19 October.
