Russian pranksters fool OpenAI’s national security team into discussing military AI
Three senior officials at OpenAI, the company behind ChatGPT, ended up discussing military artificial intelligence with two Russian pranksters after apparently mistaking them for a senior Ukrainian intelligence official. The 17-minute video call, which was later uploaded online, has raised some uncomfortable questions about security checks at one of the world’s biggest AI companies. The incident was first reported by US investigative news outlet The Intercept on November 11, 2026.
The men behind the prank were Vladimir Kuznetsov and Alexei Stolyarov, better known as Vovan and Lexus. They pretended to represent Rustem Umerov, the head of Ukraine’s Foreign Intelligence Service. The rather awkward part? OpenAI’s officials had years of experience dealing with national security matters, yet the supposed Ukrainian official reportedly kept his camera switched off throughout the conversation.
How did Russian pranksters trick OpenAI officials?
According to The Intercept, the pranksters first contacted a senior Ukrainian official in San Francisco, pretending to be Umerov. That official was already working to establish connections between Ukraine and OpenAI and reportedly helped arrange the meeting.
Once the call began, three OpenAI officials joined the discussion, apparently believing they were speaking with Ukraine’s intelligence chief.
The conversation lasted around 17 minutes. The pranksters recorded it and later published the footage on Rumble.
For a company developing advanced artificial intelligence, the deception appears to have involved something surprisingly ordinary. A false identity, a trusted introduction and a video call with the camera turned off.
:
Who were the three OpenAI officials on the call?
The meeting involved Sasha Baker, OpenAI’s head of national security policy, Caroline Zier, who leads strategic initiatives, and Ben Crampton, who handles European national security matters.
Baker and Zier previously held senior positions at the US Department of Defense. Crampton has worked with the European Union and United Nations.
The officials were discussing potential cooperation with Ukraine, particularly the use of AI in defence and military operations.
:
What did OpenAI officials reveal about military AI?
According to The Intercept’s account of the recording, the conversation covered possible military applications of OpenAI technology, including offensive capabilities.
Crampton acknowledged that some senior company leaders were reluctant to support offensive military applications. He also discussed the difficulty of separating offensive and defensive uses of AI in certain situations.
The officials mentioned contacts with the Pentagon and Ukraine’s Ministry of Defense. They also discussed Daybreak Blue, an AI system intended to help identify and fix computer security vulnerabilities.
The conversation revealed interest in closer cooperation with Ukraine. It did not establish that OpenAI had agreed to supply offensive weapons technology.
:
Who are Vovan and Lexus, the Russian pranksters?
Vovan and Lexus have been carrying out impersonation pranks since 2011. Their previous targets include British politicians David Cameron and Ben Wallace, Italian Prime Minister Giorgia Meloni and singer Elton John.
The duo frequently impersonates political leaders and government officials to obtain private conversations, which they later release publicly.
Their activities have attracted allegations of links to Russian state interests, which they deny. Their recordings have previously faced questions over editing and selective presentation.
:
What has OpenAI said about the incident?
OpenAI spokesperson Nate Evans confirmed to The Intercept that the company investigated the call weeks before the report appeared.
OpenAI said it worked with relevant authorities, established that the meeting was a prank and took steps to prevent similar incidents.
The company has not publicly detailed those measures or identified any classified information that was compromised.