Safety and security
OpenAI engineer accused of bugging company leaders, and no one says why
US prosecutors accuse an OpenAI engineer of hiding a microphone in the executives' meeting room. OpenAI calls it "sophisticated," but it sells for $13.99.
The essentials
2 confirmed facts · 2 according to the source · 3 open questions
- Federal prosecutors accuse Abdulrahman Alfozan, an OpenAI engineer, of using a device to intercept conversations. If convicted, he faces up to five years in prison.12
- The microphone was hidden in toilet paper and plugged into a TV in an executives' meeting room. OpenAI found it on September 30 during a security check.12
- According to the sourceAccording to WIRED, the device OpenAI calls "sophisticated" is a thumb-sized circuit board that sells for $13.99.2
- According to the sourceOpenAI says it has no sign that the employee accessed its models, its code or customer data.2
Why it matters
AI companies hold secrets that are worth a lot of money and that governments see as strategic. The case suggests the risk can come from inside, using cheap equipment. FBI Director Kash Patel says the engineer was after conversations about OpenAI's "superintelligence" technology. But neither the prosecutors nor the company have explained a motive2.
The details
According to the complaint reported by WIRED, Abdulrahman Alfozan spent several weeks consulting an AI assistant and a chatbot to plan the eavesdropping2. He used an internal OpenAI tool to find out which meetings were held in the room. He also used his partner's chatbot account to answer technical questions2. Only executives and their assistants can book that room, and it is where strategy, products and relations with governments are discussed2.
On October 5, OpenAI's security team found a second microphone in a bag Alfozan had left behind in the printer room2. According to prosecutors, he later picked it up from lost and found. He said a card, some cash and a pair of headphones were missing, but did not mention the device2. OpenAI says it used its own AI models to understand the microphone and work out who had placed it1.
What we don't know
- What the microphone actually recorded: the data was encrypted and OpenAI has not been able to read it.
- Why he did it, and whether he was acting alone or for someone else.
- What Alfozan says: WIRED has not been able to reach him.
Sources
- 1OpenAI employee arrested for allegedly planting a bug in a conference room
- 2An OpenAI Employee Allegedly Used a Chatbot to Help Plant a Recording Device at the Office
Was this story useful? Yes Not really