How It All Started
First, the attackers created a new WhatsApp number and pretended to be CEO Benedetto Vigna [1]. They sent a number of messages building urgency and context.
The script for the attack followed the usual line of CEO fraud: a unique business opportunity, an urgent currency transaction, a plea for discretion. To make the scenario sound plausible, the attackers used the pretext of potential problems in relations with the Chinese side.
Then came the voice call. The voice sounded absolutely identical to that of the CEO [2]. However, the recipient suspected something wrong with the scenario from the very beginning. But he decided to try a trick: asking about the book his boss recommended recently.
Fraudsters couldn’t give him an answer and hung up.
How Ferrari Saved Themselves: By Asking About One Book
Earlier, CEO Benedetto Vigna recommended a book Decalogue of Complexity by Alberto Felice De Toni to his subordinate. It wasn’t public information. The fraudster couldn’t get his hands on it. This small piece of confidential information helped the subordinate identify the fraud.
And here’s a question Bloomberg posed after hearing the story of Ferrari employee’s encounter with a deepfake attacker: How many Ferrari managers in the same situation would know what to do? How many would have such a piece of information that could never be obtained by the attacker [3]?
Why Ferrari is Important: New Example of C-Level Security Problems
Ferrari case became an example quoted in the discussion of CEO’s protection issues. Not because the attack was successful, but because the fraudster was very close to succeeding.
If you can use publicly available materials and cloning software to successfully impersonate the CEO of one of the best-known luxury brand in the world, whose every interview is public and widely discussed, then it becomes technically feasible to clone any CEO voice anywhere in the world today.
Not the question of whether someone will target your CEO using deepfake technology. But will his staff be able to recognize the fraud.
“The voice sounds like my boss. The language, the tone, everything sounds right. Until it doesn’t. And in the middle of that voice that wants so badly to be true, your brain says: ‘I think you’re lying to me.’”
— MIT Sloan Management Review, analysis of the Ferrari case
How SYNHAWK Detects This Type of Attack
SYNHAWK PROTECTION: If the Ferrari employee had access to the HAWK 7 module of SYNHAWK – an audio model that analyzes whether the voice is AI generated – he wouldn’t have needed to guess about the source of the voice. HAWK 7 would automatically recognize signs of AI generation of the voice and notify the employee immediately, even before the manager started thinking about personal questions.