Russian pranksters trick OpenAI officials into discussing military AI

What happened

Three senior OpenAI national security officials participated in a 17-minute video call with Russian pranksters Vovan and Lexus, who pretended to be Ukraine’s intelligence chief. During the call, the officials discussed topics including offensive military AI and possible cooperation with Ukraine. The incident became public on November 11, 2026, after a video of the call was posted online. OpenAI has confirmed the breach and stated that it has since improved its security protocols.

Key details

• The three OpenAI officials involved were Sasha Baker, Caroline Zier, and Ben Crampton.
• The call lasted 17 minutes and was recorded and uploaded to Rumble.
• Topics discussed included OpenAI’s Daybreak Blue AI model and potential military uses of AI.
• OpenAI has Pentagon contracts and acknowledged the event after an internal investigation.

Why it matters

OpenAI’s national security team and its partners now face increased scrutiny over information security, especially in sensitive discussions related to military AI. The company has had to implement new security measures to prevent similar incidents, affecting how it handles external communications and verification. Organizations working with OpenAI, including government agencies, may need to reassess their protocols for sharing sensitive information.