Ex-OpenAI Engineer Held Over Secret Recorder, Saudi Ties

4-minute read

Source: Tomasz Śmigla via Getty Images

A US judge agreed to keep a former OpenAI employee in custody. Prosecutors say he secretly recorded company conversations and argue that he might flee the country. They also presented evidence that he had extensive contact with a member of the Saudi royal family.

On October 9, US authorities arrested Abdulrahman Alfozan, 35, and charged him with illegally recording private conversations. Prosecutors say Alfozan placed a recording device in an executive meeting room at one of OpenAI's offices in San Francisco. OpenAI security staff later found a second, similar device in a bag he had left unattended at the company. Alfozan joined OpenAI as an engineer in July 2025. The company fired him after his arrest on October 9, 2026.

According to the criminal complaint filed on October 8, the device in the executive meeting room had been installed by at least September 28. It recorded conversations for an unknown length of time, continuing at least until October 5.

US Prosecutors Point to Saudi Arabia Ties in Request to Keep Him in Custody

On October 18, the US government asked a court to keep Alfozan, a Saudi Arabian citizen, in custody because officials feared he might flee. In a court filing, prosecutors said Alfozan had "considerable resources and ties to Saudi Arabia." They also argued that he was only days away from leaving the United States for Saudi Arabia when he was arrested. They based this claim on plans he made to take parental leave, documents about travel to Riyadh using a one-way plane ticket, and the fact that his wife and infant son had left for Saudi Arabia shortly before his arrest.

Prosecutors also say Alfozan was in frequent contact with an unnamed member of the Saudi royal family throughout 2025 and 2026, including as recently as October 5. According to reports, the royal family member told Alfozan that Saudi Arabia was starting an AI program and wanted him to join it.

"In May 2026, the individual told the defendant, 'Saudi needs you more then [sic] the US!' to which the defendant responded '100,' indicating he agreed," the court document said. "The individual repeatedly tried to recruit the defendant and asked when he planned to leave OpenAI. At the time, the defendant said he wouldn't leave until his stock vested."

US prosecutors also say Alfozan was looking into other job opportunities in Saudi Arabia before his arrest. They also say he appeared to be searching internal company data for information about OpenAI's interest in Saudi Arabia. On Monday, a federal magistrate judge in the US District Court for the Northern District of California, San Francisco Division, approved the prosecutors' request to keep him in custody while the case moves forward.

The current charge is a felony for using a device to secretly record private conversations. If convicted, he could face up to five years in prison.

"In the days surrounding the time period that the device was installed, OpenAI executives met with government officials in that very room," the October 18 filing said. "Defendant's conduct could pose a danger to the national security of the United States."

The Risk of Insider Threats in AI Companies

Insider threats, in which people inside an organization may misuse their access, are not unusual. But the allegations that this engineer secretly recorded executive meetings while staying in contact with a member of the Saudi royal family make this case especially unusual. The Wall Street Journal recently reported that OpenAI was giving employees security training similar to the training used to protect against spies. The training covered risks such as "honey pot" schemes, bribery, and secret recording. The report said the company was concerned that foreign agents might try to steal AI secrets using spying methods.

The race to develop AI involves governments and private companies around the world. Because of this, it is reasonable to expect that people may try to obtain valuable business secrets and intellectual property from companies building the most advanced large language models (LLMs), a type of AI that works with language.

Mark Beccue, chief analyst at Omdia, which is also owned by Informa TechTarget, told Dark Reading that companies such as OpenAI, Anthropic, Mistral, Google, AWS, Microsoft, and others are at the front of what he called "AI's warp-speed innovation." He said he would not be surprised if more countries were trying to recruit people inside these companies, in a way similar to what US prosecutors suggest Saudi Arabia may have tried to do with Alfozan.

"They are building and iterating the most advanced AI models, several steps ahead of open source models," he said. "All of these players, with the exception of Mistral, are US companies. Sovereign AI, the idea of developing AI within one's own country, has taken root across the globe and countries see model innovation as geo-politically strategic and key to economic growth."

OpenAI did not respond to Dark Reading's request for comment.

About the Author