The FBI arrested a Saudi OpenAI engineer over alleged bugging

The FBI arrested an OpenAI engineer for allegedly bugging a conference room at the company’s San Francisco offices, raising the specter of sensitive technology getting into the hands of US adversaries or criminals. Abdulrahman Alfozan, 35, is a Saudi national.

Who’s in charge? The case highlights how frontier AI technology is simultaneously driving the US and global economies, creating catastrophic risks if it's mismanaged. The industry is almost entirely unregulated by governments — particularly in the US — leaving private companies largely responsible for securing their models against theft or spying. (China and the European Union have taken a more hands-on approach to regulation, but don’t have as much sway over the industry given the most advanced model makers are based in the US.)

Trump has said he prefers “tremendous self-regulation.” Last month, he got OpenAI and other top AI firms to sign a voluntary pact to police themselves.

The details: Alfozan allegedly hid an encrypted recorder wrapped in toilet paper behind a TV in the conference room, the US Department of Justice says. OpenAI’s bug-sweeping team found it on 30 September, and a second device turned up days later in a bag he’d left behind. There’s no sign that Alfozan took anything from OpenAI’s systems, the company says.

The DOJ didn’t share any information about a possible motive. There’s a lengthy record of similar espionage by foreign governments and business rivals:

• A former Google engineer was convicted in January of stealing thousands of pages of Google’s AI trade secrets to benefit China.
• Last month, OpenAI said people linked to Chinese AI startup Moonshot tried to copy how its models reason.
• In 2022, a former Twitter manager was convicted of passing Saudi dissidents’ personal data to Saudi officials in exchange for cash and a luxury watch, The New York Times reported.