
An advanced recording device was found at OpenAI headquarters. The employee suspected of installing the equipment was fired, and the case is being investigated by the FBI.
AI-generated summary
OpenAI detected a recording device in the conference room and notified the FBI, using its own AI models to identify the perpetrator.
The BBC writes about the case.
According to the prosecutor's office, the device was intended to be used to intercept conversations held in a conference room intended for management staff. The equipment, wrapped in toilet paper, was hidden in a USB port on the back of the TV. It was found on September 30 - two days after it was supposed to be installed.
OpenAI said it detected an "advanced recording device" and notified the FBI. The company said it used its own AI models to determine that an employee had installed the device. Alfozan was released.
Investigators indicate that the suspect allegedly used tools based on chatbots and productivity applications. In this way, he was supposed to find information about places at the company's headquarters and about the construction, masking and remote operation of sound recording equipment.
On October 5, Alfozan was scheduled to return to the room to inspect the device. The prosecutor's office claims that his presence in this place is confirmed by surveillance recordings. On the same day, OpenAI security found a bag with his documents, wallet and a second, similar recording device left in another part of the building.
The company assured that there was no evidence of unauthorized access to the source code or customer data.
Alfozan appeared in court on Tuesday. If found guilty, he faces up to five years in prison and a fine of up to PLN 250,000. dollars – writes the BBC.
AI outlook — possibilities, not facts
Legal proceedings against Alfozan
Very likely · Within months

CSIRT KNF warns Android users against a fake application that steals payment card data using the NFC Relay method. The fraud spreads, among others: via SMS and fake bank websites.

Research involving 1,222 people showed that using AI to solve tasks weakens perseverance and the ability to think independently. Participants supported by ChatGPT performed worse after being removed from the tool than those working alone.

Following the revelations of incidents involving the Claude model, such as bogus visa applications and a false homicide report, the White House ordered AI companies to immediately report security breaches and take corrective action, emphasizing that this is a national security responsibility.

The Accessibility Discovery Center was opened in Warsaw, where Google engineers cooperate with people with disabilities. The center offers tests of AI technologies, eye control and tools improving the accessibility of mobile devices.

ArXiv, the electronic archive of scientific preprints, is experiencing a surge in submissions due to how easily AI can generate articles. In September 2026, the number of submitted papers reached a record 40,363, overloading moderators and making it difficult to find valuable research. Mathematicians such as Kevin Buzzard and Terence Tao warn that many AI-generated articles have low content and are difficult to understand, despite sometimes correct solutions. In response, arXiv has introduced publication limits (two articles per month, three active submissions) and a requirement to label AI use, while supporting initiatives like Hexagon that collect “pre-preprints” requiring further review.

Italian Prime Minister Giorgia Meloni has registered her voice as a sound trademark with EUIPO. The goal is to protect against artificial intelligence cloning and illegal use of the famous words of her speech.