A federal court on Monday refused to release Abdulrahman Alfozan, a former OpenAI engineer accused of bugging a top-secret conference room to steal the company's secrets.
The ruling granted a request from US attorney Craig Missakian, who argued in a court filing Sunday that Alfozan must stay detained while the government builds its case. Prosecutors say he spied either for his own business gain or to benefit the Kingdom of Saudi Arabia.
The government's evidence, as it described it, has three parts. Alfozan had been chatting on WhatsApp with a member of the Saudi royal family. He was also pursuing opportunities to move back to Saudi Arabia permanently, during what prosecutors called the country's biggest AI push yet. And shortly before his arrest, the FBI found he had opened a folder called "Saudi Gov Docs" on his iPhone, where he also allegedly kept photos of a second bug he planted in the offices. No response from Alfozan or OpenAI was available.
A detention ruling is not a verdict, and the allegations are untested at trial. The pieces prosecutors cite are also mostly links and labels: a chat partner, a job search, a folder name. A folder title shows how someone sorted files, not necessarily what was in them. The photos of a second bug are the most concrete item in the description, but the court has not weighed them as proof of guilt.
The case fits a pattern. Frontier AI labs hold model designs and research that governments and rivals want, and prosecutors have already charged engineers over AI trade secrets, including a former Google engineer indicted in 2024. Those cases mostly involved copying files an employee could already reach. A planted bug in a conference room, if proven, would be a more physical kind of spying than anything in that precedent.