A senior Ukrainian official based in San Francisco reportedly helped Russian pranksters obtain a video meeting with three OpenAI leaders overseeing national security and strategic initiatives. According to The Intercept, the call reportedly ran for 17 minutes, covering military uses of AI and ongoing discussions with the US Department of Defense about supplying Ukraine with AI tools.
OpenAI’s participants were Sasha Baker, Caroline Zier, and Ben Crampton. Their callers claimed to be senior Ukrainian government officials but were actually Russian comedians Vovan and Lexus, respectively Vladimir Aleksandrovich Kuznetsov and Aleksei Vladimirovich Stolyarov. The pair recorded the meeting, then published the recording online.
Vovan described arranging the call as surprisingly easy in his account to The Intercept. He said the comedians contacted a senior Ukrainian official working in San Francisco, who offered to set up a meeting with OpenAI. In other words, the pranksters got in through a trusted introduction, not by breaking into the company’s computer systems.
Asked about offensive military AI services for Ukraine, Crampton said he wanted stronger ties but acknowledged limits on how fast they could grow. Responding to a question about offensive military AI services for Ukraine, Crampton expressed a desire for stronger ties while acknowledging limits on how fast they could grow.
“There’s no question, I think, from our perspective that we want to have as deep a relationship as we possibly can with Ukraine,” Crampton is heard saying. “But it may not be possible to get to where we want to be straight away.”
He also discussed resistance within the company to offensive uses of OpenAI technology, suggesting that addressing executives’ reservations formed part of the national-security staff’s job.
“Many of the company’s leaders would not want our technologies to be used in offensive military operations,” Crampton said on the call. “Our task is to explain that in certain situations, the line between offense and defense can be quite conditional.”
Those remarks do not settle the policy question of whether aid framed as supporting Ukraine’s defense can remain distinct from offensive military applications. Nor do the comments, by themselves, establish that OpenAI agreed to provide offensive capabilities.
Crampton further described an OpenAI connection to Ukraine’s Ministry of Defense, “primarily through a colleague of ours based in Washington… who previously had worked with the Delta team in your Ministry of Defense.”
Futurism argues that the episode raises serious doubts about OpenAI’s security practices: if an introduction can bring pranksters into contact with senior staff, malicious actors might gain similar access. The publication places that concern alongside OpenAI’s large Pentagon contract to deploy “advanced AI systems in classified environments.” Still, nothing in the incident shows those classified environments were compromised.
Responding to The Intercept, OpenAI said that “weeks ago, we launched a thorough investigation into the call’s authenticity.” Its spokesperson added: “We acted internally and with the relevant authorities to conclude that it was a prank conducted by individuals with a history of targeting officials and other companies.” Vovan and Lexus have pranked public figures before, including John McCain and Bernie Sanders.
“We took action to prevent similar situations from happening in the future,” the OpenAI spokesperson said. OpenAI did not specify whether it had changed how it verifies callers or handles introductions to employees working on national security.
