OpenAI says its own AI helped catch an engineer who hid a recording device in an executive meeting room

OpenAI says it found a recording device hidden in one of its offices, used its own AI models to work out where it had come from, and traced it to one of its engineers, who has since been fired and arrested by the FBI. The company announced the arrest on X on Tuesday evening, saying the employee had planted the device “a few days earlier”.

We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work.

At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated.

We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.

— OpenAI Newsroom (@OpenAINewsroom) October 13, 2026

OpenAI didn’t name the employee or say which office was targeted. It said the FBI’s investigation is ongoing.

Hidden in toilet paper behind a TV

Fox News correspondent Bill Melugin posted an excerpt from the criminal complaint shortly afterwards. He named the man as Abdulrahman Alfozan, a Saudi national, and said he made his first appearance in federal court on Tuesday. The complaint refers to OpenAI only as “Company 1” and says Alfozan had worked there as an engineer since July 2025.

According to the excerpt, OpenAI staff found the device on or about September 30, “inserted into a USB port on the back of a television in an executive conference room” in one of its San Francisco buildings. It had been “concealed in toilet paper”, was installed on or about September 28, and its contents were encrypted, the complaint says.

Images from the complaint that Melugin shared show a small circuit board branded Seeed Studio, a maker of boards for hobbyist electronics, and a still from a motion-activated camera behind the screen, timestamped October 5, days after the device was found. Melugin says Alfozan was caught on OpenAI’s security cameras.

The Justice Department hasn’t published a press release, and the charges haven’t been made public. Alfozan hasn’t commented, and it isn’t known whether he has a lawyer. Nothing made public so far says who the recordings were meant for.

“No indications” anything was taken

OpenAI says it has “no indications” that the engineer accessed or took model weights, source code or customer data. It hasn’t said whether the device captured any conversations, or what was discussed in the room while it was there.

The case lands in a tense few weeks for OpenAI over who can get at its information. On October 1 it let three safety researchers go for allegedly sharing company information with an outside safety group, and the day before it accused people linked to China’s Moonshot AI of trying to copy its models’ hidden reasoning.

AI as the detective

OpenAI presents the case as proof of “how AI can help address these sophisticated security threats”, but says little about what its models did, beyond helping it “understand the device and where it had come from”. It hasn’t said which model it used, and the complaint’s mention of camera footage suggests ordinary security tools played a part too.

The pitch fits a wider argument in the industry about insiders. Microsoft’s Satya Nadella said last weekend that companies should treat AI models themselves as insider risks; here, OpenAI says a model helped catch a human one.

Why it matters

AI labs guard model weights and plans worth billions, and a conversation in an executive meeting room can be as valuable as code. OpenAI has framed this as a win for AI-powered security, but the biggest questions are still open: what the device recorded, who it was for and what the engineer is charged with.

Sources: OpenAI Newsroom on X; Bill Melugin, Fox News, on X.