OpenAI Executives Pranked by Russian Comedians Discussing Military Support for Ukraine

Russian Pranksters Fool OpenAI Security Officials

Three of OpenAI's highest-ranking national security staff fell victim to an elaborate impersonation scheme orchestrated by Russian pranksters known as Vovan and Lexus. The OpenAI executives pranked included Sasha Baker, head of national security policy, Caroline Zier, lead for strategic initiatives, and Ben Crampton, lead for European national security initiatives. The trio joined what they believed was a legitimate video call with Rustem Umerov, secretary of Ukraine's National Security and Defense Council, to discuss potential collaboration between OpenAI and Ukrainian forces1

1

2

Source: ET

The 17-minute conversation, later reported by The Intercept, revealed how Russian pranksters successfully bypassed OpenAI's security protocols by first approaching a senior Ukrainian official in San Francisco while impersonating Umerov. The Ukrainian official, believing the request was genuine, facilitated an introduction to OpenAI's national security team. During the call, the impersonator kept his camera switched off throughout, while Baker and Zier both stated they had met Umerov previously and expressed pleasure at speaking with him again2

2

Intelligence Discussions on Military AI Applications

Source: Digit

The OpenAI security officials fooled by the scheme engaged in detailed intelligence discussions about military support for Ukraine and the potential deployment of AI technology in combat scenarios. According to the recording, Crampton revealed that many senior OpenAI leaders remained hesitant about allowing the company's technology to be used for offensive military operations. Despite this internal resistance, he expressed interest in building a closer relationship with Ukraine and exploring ways the company could support Ukrainian defense efforts2

2

The conversation touched on geopolitical topics including cooperation with Ukraine's Ministry of Defense and discussions about OpenAI's Daybreak Blue model, an AI system designed to identify and fix computer vulnerabilities. The executives also discussed sensitive conversations with the Pentagon regarding support for Ukraine, revealing the depth of OpenAI's engagement with military and intelligence communities on these matters1

1

2

OpenAI Confirms Breach and Pledges Stronger Safeguards

OpenAI confirmed the incident after launching an investigation weeks before The Intercept's report was published. In a statement, the company acknowledged that it worked internally and with relevant authorities to verify the call's authenticity and concluded the meeting was indeed a prank. The AI giant pledged to implement stronger safeguards to prevent similar security vulnerabilities in the future, though it did not disclose the specific security measures introduced or the exact date of the meeting2

2

This incident raises concerns about how easily bad actors can exploit trust relationships and social engineering tactics to access sensitive information from technology companies working on national security matters. The breach highlights the challenges AI companies face when balancing collaboration with government partners against the need for rigorous verification protocols, especially when dealing with geopolitical topics and military applications of their technology.

Related Stories

OpenAI Faces Major Security Breach and Ethical Concerns

OpenAI Pauses AI Model Training After Agent Hacks Australian Government Website

Security Researchers Used Anthropic Claude to Breach OpenAI in $6,500 Bug Bounty Program

OpenAI Agents Escape Sandbox, Post 18,000 Messages on German Wiki to Share Hacking Techniques

Who Are Vovan and Lexus?

Vovan and Lexus are the pseudonyms of Russian pranksters Vladimir Kuznetsov and Alexei Stolyarov, who have built a reputation for impersonating public figures since 2011. Their previous targets have included musician Elton John and the late US senator John McCain, demonstrating a pattern of targeting high-profile Western figures. Both pranksters have faced allegations of links to Russian intelligence and government circles, though they have consistently denied working for the Russian government2

2

Source: Futurism

The sophistication of this latest operation suggests an evolving threat landscape where state-adjacent actors can exploit the complex web of international relationships in the AI sector. Watch for increased scrutiny of how AI companies verify the identity of government officials and whether this incident prompts industry-wide changes to authentication protocols for sensitive discussions involving military and intelligence partnerships.