Three senior OpenAI officials were reportedly tricked into a 17-minute video call by Russian pranksters, who posed as a Ukrainian government representative. During the deceptive call, the officials discussed the company's potential military AI support for Ukraine, including sensitive offensive capabilities, according to reports from The Intercept.
The elaborate prank was orchestrated by notorious Russian duo Vovan and Lexus, who impersonated Rustem Umerov, head of Ukraine’s Foreign Intelligence Service. A recording of the conversation was subsequently posted online, revealing the extent of the deception.
OpenAI Officials Discuss Military AI Capabilities
The OpenAI representatives on the call included Sasha Baker, who oversees national security policy; Caroline Zier, lead for strategic initiatives; and Ben Crampton, who manages European national security matters for the company. The fake Ukrainian official probed OpenAI's stance on providing military support and whether it could offer offensive military AI services.
Ben Crampton expressed OpenAI’s desire for a strong relationship with Ukraine but acknowledged internal reluctance regarding offensive applications of their technology. “There’s a general reluctance on the part of many people who are senior in the company to get too involved in offensive uses of our technology in a military context,” Crampton stated. He also highlighted the challenge of distinguishing between offensive and defensive uses in certain scenarios.
Sasha Baker confirmed that OpenAI had been in contact with the Pentagon concerning assistance for Ukraine. The officials also mentioned discussions around Daybreak Blue, an OpenAI AI model designed to identify computer vulnerabilities, and a burgeoning partnership with Ukraine’s Ministry of Defense.
Company Response and Pranksters' History
Following the incident, OpenAI launched an internal investigation. Company spokesperson Nate Evans confirmed the call was a prank by individuals known for targeting public figures and organizations. OpenAI stated it took action to prevent similar occurrences in the future.
Vovan and Lexus have a history of successful prank calls dating back to 2011, with past targets including Elton John and John McCain. Their pranks frequently align with Russian foreign policy interests, though they deny direct government affiliation.
The discussion with the pranksters also briefly touched on unrelated topics, such as concerns about Elon Musk’s influence and former U.S. President Donald Trump’s stance on the conflict. Baker noted that OpenAI could not assist with issues related to Musk, humorously adding, “He’s not a fan of our CEO either, if it helps you at all. So, we’re in the same boat.” Regarding Trump, Baker expressed that while his comments were unfortunate, the majority of the U.S. government and public remained supportive of Ukraine.
This incident comes amid OpenAI’s existing $200 million contract with the U.S. Department of War, which focuses on providing AI applications to “warfighters” to enhance military advantage and decision-making.