OpenAI says it found a hidden recording device planted in one of its executive conference rooms, and the person charged with putting it there is one of its own employees.

According to OpenAI, the device was discovered in its San Francisco office about a week before the FBI arrested product-team employee Abdulrahman Alfozan on October 9. He had joined the company in July 2025 and has since been fired. Federal prosecutors charged him with illegally intercepting oral communications, a crime that carries up to five years in prison. He has not been convicted, and neither he nor his attorney has commented publicly.

The setup was low-tech in placement and high-tech inside. The device was plugged into a USB port on the back of a TV in the conference room. OpenAI says it then used its own AI models to analyze what it had found, describing a device with strong encryption, audio monitoring, and a way to move data out while avoiding detection. Prosecutors say a camera OpenAI placed near the TV caught Alfozan checking on it.

What it was for is still unknown. The FBI says the aim was to capture sensitive conversations about OpenAI’s most advanced AI work, and the bureau is treating the case as a counterintelligence priority.

The reassuring part: OpenAI says it has seen no sign that the employee accessed or took model weights, source code, or customer data.

Thankfully the breach appears to have been caught early, and the crown jewels look untouched. But the threat to frontier AI labs isn’t only hackers on the outside. It’s people with a badge walking past the conference room every day.

As these models get more valuable, the pressure to steal what’s said about them in private rooms will only grow, and labs will have to guard their people and offices as closely as their code.

How worried?: 🟡 3/5 — Meaningful