Russian pranksters trick OpenAI executives into discussing Ukraine military AI

The high-level prank

A Russian internet comedy duo known as Vovan and Lexus successfully deceived three of the top military and national security leaders at OpenAI. The pranksters pretended to be representatives from the Ukrainian government to start a candid talk about providing weapons to Ukraine. The group has a long history of making prank calls to famous figures, including Elton John and John McCain, dating back to 2011.

The victims of this specific prank included Sasha Baker, who runs national security policy at OpenAI; Caroline Zier, the company’s lead for strategic initiatives; and Ben Crampton, who leads the company’s European national security matters. All three individuals have extensive government experience, with Baker and Zier previously holding high-level postings in the U.S. Department of Defense. Crampton has worked for both the European Union and the United Nations.

The pranksters arranged the meeting by pretending to be Rustem Umerov, the head of Ukraine’s Foreign Intelligence Service. During a 17-minute video conversation, the OpenAI officials discussed the company's military support to Ukraine. The Russian group later uploaded a recording of the meeting to the platform Rumble to share the contents with the public.

Details of the conversation

During the call, the OpenAI executives discussed the possibility of providing offensive capabilities to aid Ukraine’s defense against Russia. Ben Crampton stated that while the company wants a deep relationship with Ukraine, it might not be possible to reach their goals immediately. He noted that senior leaders at OpenAI are generally reluctant to get involved in the offensive use of technology in a military context.

The conversation also touched on OpenAI's ongoing work with the Ukrainian government regarding its Daybreak Blue AI model. This specific system is designed for fixing computer vulnerabilities. Crampton also mentioned a nascent partnership with the Ukrainian Ministry of Defense, noting that they are interested in partnering in as many ways as possible.

The executives also shared information regarding OpenAI's $200 million contract with the U.S. Department of War. According to contract documents, this deal provides AI applications for warfighters to improve military advantage and decision-making. The participants even discussed political figures, including Donald Trump and Elon Musk, with Baker noting that the company is in the same boat as Ukraine regarding Musk's stance.

How the trick worked

Vladimir Kuznetsov, a member of the Russian comedy duo, explained that the prank was relatively simple to organize. They initially contacted a senior Ukrainian official located in San Francisco while posing as Mr. Umerov. This official was attempting to establish connections with OpenAI through the Ukrainian diaspora in California.

The official in San Francisco was already in contact with Sasha Baker because of her previous ties to Ukraine during her tenure at the U.S. Department of Defense. Believing he was speaking directly with Umerov, the San Francisco official offered to arrange the meeting with the OpenAI team. The pranksters were able to leverage these existing professional connections to gain access to the high-level executives.

During the video call, the faux-Umerov kept his camera turned off for the entire 17-minute duration. This allowed the pranksters to maintain their disguise while the OpenAI officials spoke freely. The executives were so convinced of the identity that they even commented on having met Umerov previously and expressed pleasure at speaking to him again.

OpenAI's response

OpenAI spokesperson Nate Evans released a statement to The Intercept confirming that the company launched a thorough investigation into the call's authenticity. The company concluded that the interaction was a prank conducted by individuals with a history of targeting officials and other companies. They stated that they acted internally and with relevant authorities to reach this conclusion.

The company confirmed they have taken specific actions to prevent similar situations from happening in the future. While the Ukrainian Embassy in Washington did not immediately respond to requests for comment, the incident highlights a significant security lapse. It shows that even high-ranking officials with deep government experience can be deceived by social engineering tactics.

The incident serves as a reminder of the risks associated with social engineering in high-stakes environments. It highlights how easily established trust and professional networks can be exploited to extract sensitive information. As AI companies continue to engage with global governments, these types of security vulnerabilities remain a critical concern for the industry.