US authorities have arrested an engineer working at OpenAI after an audio recording device was found hidden behind a television in an executive meeting room at the company’s headquarters in San Francisco. Prosecutors also revealed that an account linked to his wife was used to ask a chatbot for help building and concealing a recording device.
اضافة اعلان
The engineer was arrested after a listening device was discovered inside the company’s San Francisco office. He has been charged with attempting to unlawfully intercept oral communications. OpenAI said there were no indications that its AI model weights, source code or customer data had been stolen.
According to the US Department of Justice, the suspect, Abdulrahman Al-Fawzan, 35, is a Saudi-born national who has worked at OpenAI since July 2025. He was arrested on October 9 and appeared before a federal court on Tuesday, October 13, 2026.
Prosecutors said the recording device was connected to a USB port behind a television in a meeting room reserved for senior company officials. It was allegedly placed there around September 28 and discovered by employees two days later.
The device was wrapped in toilet paper, and its contents were encrypted. On October 5, a second device was found inside a bag the engineer had left unattended in another room.
Chatbot use cited in the investigation
Among the allegations disclosed by prosecutors was that an account linked to the engineer’s wife, sharing some identifiers with his own account, was used to send text prompts to a chatbot requesting help with building and concealing an audio recording device in the days leading up to the incident.
Prosecutors also said surveillance footage showed the engineer near a meeting room he was not authorized to enter.
Al-Fawzan faces a potential sentence of up to five years in prison and a $250,000 fine if convicted. He remains accused, and no conviction has been issued.
OpenAI denies data theft
The case comes amid a separate dispute involving OpenAI. Apple has accused former employees of bringing designs and prototypes to job interviews at OpenAI, allegations that OpenAI denies.
In the listening-device case, prosecutors have not accused any foreign government of directing or instructing the engineer. Although US officials have spoken broadly about foreign threats and the need to protect American technology, they have not identified any country as being involved in the case.
OpenAI previously faced a security incident in July, when some of its models left an isolated testing environment during an internal evaluation. Researchers described the incident as a failure to contain the models.
Regarding the current case, the company said it used its ownAI models to understand the nature of the recording device and identify the person who planted it.
In a statement carried by Bloomberg, the company reiterated that there was no evidence that model weights, source code or customer data had been taken.
Questions over European security obligations
The incident raises questions about OpenAI’s regulatory obligations in the European Union, following the company’s signing of the European AI Code of Practice. The code includes requirements to report certain security incidents to the AI Office within five days.
However, it remains unclear whether the listening-device incident falls within the category of events that must be reported. These requirements concern serious incidents involving models placed on the market, while OpenAI maintains that none of its models was compromised.
The case also comes as OpenAI works to establish a European headquarters in Dublin, Ireland, where privacy engineers will operate under the supervision of Irish regulators.
Source: Al-Ghad.
