OpenAI Employee Uses "Unnamed Chatbot" to Spy on OpenAI, Gets Arrested by FBI
OpenAI is in the news again. This time, it’s due to a recent incident involving corporate espionage. According to a statement the AI lab released on 13th October, the company discovered a “sophisticated recording device” at one of its offices. The company quickly detected it and soon realized an employee had planted the device. After discovering the attempt, the company handed the case to the FBI and thanked the (human) agents for quickly arresting their employee.
While being arrested by the FBI is rarely a happy moment, this incident feels quite funny, and it gets funnier the more you learn about it.
The employee in question is a national of the Kingdom of Saudi Arabia and is also listed as a partner at a venture capital firm focused on the Middle East and North Africa. The Department of Justice has not released any motive, but it has released a lot of information that provides a clearer picture of what transpired in this case:
- The accused has been employed at the company since July 2025.
- CCTV footage showed the employee hanging around an executive conference room on multiple days even though he had no access to the meetings taking place within that room. These meetings were reportedly highly confidential and internal to the company.
- The employee used “a productivity tool” to learn more about the company’s building addresses, meeting rooms, “leadership and exec” locations, etc. The employee’s spouse also had an account associated with an “unnamed chatbot” that contained several queries about building, installing, audio recording, testing, and several other prompts that clearly show the intention of building a remote recording device.
- On 28th September, the employee likely installed their device in the conference room. They hid it in toilet paper (?) and inserted it into a USB port on the back of a television.
- By 30th September, the gig was up, and OpenAI had discovered the recording device and started acting on the information.
- On 5th October, the employee went back to the meeting room to take a look behind the television to see if his recording device was working or not. A hidden camera placed behind the TV captured an image of him doing this.
- Additionally, the employee left a bag unattended in another room in an OpenAI building, which contained another recording device, along with his wallet and company ID.
- The FBI is currently investigating the case, and the accused can face up to five years in prison and a $250,000 fine.
So they definitely weren’t dealing with a Robert Kerbeck. The company has also made it clear that any model weights, source code, or customer data were not leaked or stolen.
So, what can we learn from this? Quite a few things, actually.
- Do not spy on your employer and reverse the expectations.
- If you do spy on them, do not ask a chatbot (which is likely owned by your employer) how to do it.
- Do not take OpenAI’s statements seriously. They claim this case is “an example of how AI can help address these sophisticated security threats.” How sophisticated can someone be when they leave their recording device and identification papers unattended in a random room at the company they are spying on? How sophisticated can someone be when they don’t even know that everything you say to any chatbot can and will be used against you? If this is the level of sophistication their security program is built for, they shouldn’t pat themselves on the back; they should aim higher.
Links referred to in this article: