
Tech • AI • Robotics • Game
OpenAI’s lead engineer for the Agents platform was arrested after prosecutors said he planted a homemade listening device in an executive conference room, where it was discovered within 48 hours.
Federal authorities arrested Abdul Raman Alphazan, 35, a Saudi national and U.S. permanent resident, on allegations that he secretly recorded conversations inside OpenAI headquarters. Prosecutors charged him with one count of federal wiretapping, a crime that can carry a sentence of up to five years in prison.
Alphazan joined OpenAI in July 2025 as an engineering lead on the product team and was described as the lead engineer for the company’s Agents platform. Before that, he had studied at Berkeley and MIT and had previously worked at Meta.
Investigators say the effort started about a month before the arrest, when Alphazan allegedly used an internal agent to gather information on meeting rooms, including room capacities, calendars, and the location and timing of leadership and executive meetings. Prosecutors say he also sought details on when outside partners were expected to visit the headquarters.
Between September 24 and September 30, prosecutors say Alphazan used his wife’s ChatGPT account to ask how to build a Wi‑Fi-enabled audio recording device with an encrypted SD card. The device described in the complaint was built around a $23 Seeed Studio microcontroller with a low-cost antenna and was allegedly wrapped in toilet paper to make it less conspicuous.
On September 28, prosecutors allege that Alphazan placed the device in the executive conference room by plugging it into a USB port on the back of a television. The target was a room used for senior leadership meetings, raising concerns that sensitive corporate discussions could have been captured.
The device remained in place for about 48 hours before OpenAI’s technical surveillance countermeasures team found it during a routine sweep. That short lifespan sharply limited any potential intelligence value and immediately triggered an internal investigation.
Investigators said the device was easy to trace. The complaint indicates that OpenAI used its own models to analyze the hardware, while a more basic clue also stood out: the board’s MAC address had allegedly been pasted into the wife’s account days earlier. Security footage also reportedly showed Alphazan near the executive room on six separate days.
After finding the bug, the company reportedly installed a hidden camera behind the room’s television and waited to see whether anyone would return. Prosecutors say Alphazan came back on October 5, strengthening the case against him. He was then fired and arrested on the same afternoon, four days later.
The criminal complaint does not identify any outside handler or employer, and the charge is wiretapping, not espionage. That distinction matters because espionage charges would require proof that the recordings were gathered on behalf of a foreign government. Public scrutiny has nevertheless focused on Alphazan’s ties to RED Ventures, a venture capital firm in Riyadh, where reporting has noted links to a Saudi prince whose father is a senior intelligence official.
The case adds to a difficult period for OpenAI. It follows other recent controversies involving internal handling of sensitive information and product safety concerns, intensifying questions about how the company secures its systems, facilities, and high-level discussions as its influence in artificial intelligence continues to grow.
The allegations describe a striking insider security breach carried out with crude hardware and an unusually visible trail. The unresolved question is not whether the device was planted, but whether the wiretapping was a lone act or part of a broader intelligence effort.
Ask a question