OpenAI, the tech giant behind ChatGPT, said it found a hidden recording device tucked into one of its executive conference rooms.
The San Francisco-based AI company said yesterday that staff had to use one of its models to ‘understand the device and where it had come from’.
The bug was slipped behind the TV of a room where high-level chats about confidential plans, government relationships and research happened.
A ‘specific employee had planted it a few days earlier’ in late September, OpenAI said on X, adding that the staffer has since been arrested.
The US Department of Justice named the worker as Abdulrahman Alfozan, 35, from Saudi Arabia.
He has been charged with using or endeavouring to use an electronic, mechanical, or other device to intercept oral communications.
Alfozan appeared in court today and remains in custody.
Between 17 August and 16 September, Alfozan is accused of asking a productivity AI assistant questions about meeting rooms.
They included ‘can you list today’s meeting in this room’, according to the court document.
Prompts – instructions to an AI – to a chatbot used by Alfozan’s spouse included how to build a ‘continuous audio recorder’.
Alfozan was seen on CCTV near the executive conference room ‘on multiple dates’ just before the gadget was planted, the complaint said.
The wiretapping device was stuffed into a USB stick, wrapped in toilet paper and planted behind a TV in the room on September 28, officials said.
OpenAI discovered the spy device two days later.
A photo in the court document shows an object on the back of a TV.
Another image shows the gadget with the branding of Seeed Studio, which makes AI-powered hardware.
Alfozan had no authority to access the meetings in the room but footage allegedly shows him peeking behind the TV and looking at where the recording device had been stashed on October 5.
OpenAI told that the recording device was designed to evade detection and used ‘advanced encryption’, so it would be tough to crack, and could monitor audio and had a way to transfer data without being noticed.
The company also has evidence that the employee intended to deploy other recording devices in top-secret rooms.
After searching a bag left unattended by Alfozan in OpenAI offices, FBI agents discovered a second recording device, according to the indictment.
If convicted, Alfozan faces five years in prison and up to $250,000 (nearly £190,000) in fines.
Alfozan began working at OpenAI in July 2025. Since September last year, he has been an engineering lead, according to a LinkedIn profile bearing his name.
OpenAI fired him on October 9, the day he was arrested. It timed the firing of its employee with the FBI to avoid disrupting the investigation.
The company gradually limited the worker’s access to company systems and offices, then suspended his access, put him on leave and confiscated his work equipment.
Alfozan’s lawyer, William Pernik, told that his client is ‘presumed innocent’.
‘These allegations appear wholly out of character for him, and we will relentlessly investigate all evidence produced by the government in this case to understand what actually happened, who was responsible and any possible reasons why,’ he added.
OpenAI said: ‘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.’
Get in touch with our news team by emailing us at webnews@metro.co.uk.
MORE: Microsoft reveals Xbox Labs and it’s desperately pretending it’s not about AI
MORE: Trump shares disturbing video of Yorkshire tiger mauling with ‘warning’ to America
MORE: Games Inbox: What video game console is best to get for Christmas?