OpenAI Engineer Charged Over Hidden Recorder in Exec Room — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Abdulrahman Alfozan, a 35-year-old Saudi Arabian national and US permanent resident, was charged with one count of intercepting oral communications and faces up to five years in prison after allegedly concealing an audio recording device inside a piece of toilet paper and inserting it behind a monitor in an OpenAI executive-only conference room in San Francisco.
- Alfozan allegedly spent several weeks using OpenAI's internal AI chatbot and "productivity agent" — including an account tied to his spouse — to research covert recording setup and to query meeting schedules for the targeted conference room.
- OpenAI security discovered the first device on September 30 during a countersurveillance sweep, then found another Seeed XIAO ESP32-S3 Sense recorder (which retails for $13.99) in an unattended bag Alfozan left in a printer room on October 5; both devices' storage was encrypted and unreviewable.
- The targeted conference room is reserved only by executives and executive assistants and is used for leadership meetings, technical and product sessions, and external briefings with business partners and government officials involving "highly confidential" information.
- OpenAI fired Alfozan last Friday to coincide with his arrest; the company says it found no indications he inappropriately accessed or exfiltrated model weights, source code, or customer data, while FBI Director Kash Patel stated Alfozan aimed to capture communications about the company's "super intelligence technology."
- Alfozan previously worked at Meta and at a cargo logistics startup, and is listed on LinkedIn as a part-time venture partner at Saudi Arabia-based RAED Ventures; prosecutors and the company have not publicly identified a motive.
Why it matters: Alfozan allegedly used OpenAI's own AI chatbot to plan the bugging of an executive-only room used for highly confidential government briefings — meaning the case exposes both an insider-threat vulnerability at a frontier-model developer and a live example of AI tools being used to facilitate corporate espionage, even though no motive or data exfiltration has been confirmed.
Ask SkimNews
