OpenAI says it investigated a call in which people posed as Ukrainian officials and concluded it was a prank. The company’s statement appeared in The Intercept’s November 11 report. The outlet identified the callers as the Russian duo known as Vovan and Lexus and said OpenAI staff who work on national security joined the call.
OpenAI spokesperson Nate Evans told The Intercept the company began looking into the call weeks earlier and worked with authorities. He said OpenAI took steps to prevent a repeat, without specifying what changed. The report gives no exact date for the call.
Test a convincing invitation
The National Institute of Standards and Technology’s phishing guidance advises checking a request through contact details you already trust. A team that handles sensitive briefings can turn that advice into a short exercise.
Have a real colleague forward a mock invitation from someone claiming to represent a partner. Give the meeting host a routine agenda and a slide with mock restricted data. Ask the host to reach the partner through a known number or directory and confirm who will attend. Have the person responsible for the data separately decide which material that guest can receive.
Run the exercise twice. In the first run, the partner cannot confirm the visitor. The host should postpone the sensitive briefing. In the second run, the partner confirms the visitor, but the data owner has cleared only the routine material. Check whether the host can open the meeting without sharing the restricted slide. This is a proposed exercise. The public record does not describe how OpenAI screened the invitation.
Abdessalam Alaoui
I'm Abdessalam Alaoui, founder of Musthave.AI. I write practical guides to AI models, tools, generative media.
