He Used ChatGPT To Plan Corporate Espionage At OpenAI. Then His Own Team Caught Him.
(SeaPRwire) –
By: Ethan Gallagher
A former OpenAI engineer allegedly used ChatGPT to plan a month-long surveillance operation inside his own company. He then reportedly used an internal agentic AI tool to execute it. Someone planted a recording device at the San Francisco headquarters. OpenAI caught them doing it. The irony is almost too clean to be real. But the federal case is real enough.
Abdulrahman Alfozan, a Saudi national and legal permanent resident since 2020, was arrested on October 12. He was trying to board a one-way flight to Saudi Arabia with his wife and young child. The flight was prepaid. OpenAI terminated him on October 9 after security found the device during a routine sweep on September 30. The device was allegedly planted on September 28, after Alfozan spent over a month planning the operation on ChatGPT and an internal AI tool, according to the federal complaint. His lawyer, William Pernik, called the allegations “wholly out of character” and argued the public should not jump to conclusions based on nationality. The next court date is October 19. Neither the FBI nor OpenAI has found evidence of leaked information. Alfozan is charged with violating a general federal recording statute, not sharing trade secrets. Pernik insists his client was entitled to due process and that the government should protect intellectual property while respecting the law.
The FBI pressed Alfozan’s Saudi nationality hard in its press release. He moved to the U.S. in 2011 to attend MIT’s computer science program. He worked at SAP, Microsoft, and Meta for five years before founding a Silicon Valley startup. He joined OpenAI in July 2025 and also became a partner at Saudi venture firm RAED Ventures that same year. Pernik pushed back on the national origin framing, noting Alfozan had lived, studied, and raised a family in the U.S. for over a decade. The government’s complaint paints a picture of calculated insider risk. The defense paints a picture of a tech worker whose plans never materialized into leaks. What sits between those two portraits matters for every AI lab watching this unfold.
The broader takeaway is uncomfortable. The same tools these companies ship to the world became the weapon in a corporate espionage plot. Alfozan used ChatGPT to plan, an internal AI agent to assist, and physical access to place hardware. OpenAI’s own security sweep stopped him. But the case exposes a structural vulnerability. Insider threat programs were built for human behavior patterns and credential monitoring. They were not built for an engineer who treats proprietary AI as a co-conspirator. The supply chain of trust inside AI labs just got a lot more complicated.
Author bio: Ethan Gallagher is a Silicon Valley hardware architect and infrastructure strategist with two decades of experience in secure system design and enterprise technology operations.