// ZDNET — INTELLIGENZA ARTIFICIALE
A low-tech solution from the past may be your best defense against AI deepfakes
Follow ZDNET: Add us as a preferred source on Google.
In January 2024, an employee at professional services firm Arup joined a video call with someone they believed included the company's CFO. This call resulted in 15 wire transfers to third-party accounts totaling about $25m.
Every participant on the other side of the video call was an AI-generated clone cobbled together from public appearances and earnings calls of Arup executives.
Also: 6 essential strategies to defend against AI-powered threat actors
"Seeing and hearing someone is no longer proof they are real," said Deepak Gupta, technical CEO at GrackerAI, when discussing the Arup incident with ZDNET. "Any protocol that relies on 'I recognized their face and voice' is now broken."
We live in a post-truth world where employees handle millions of dollars in company funds every year. Mistakes can be expensive, even when attacks don't lead to a direct loss of funds. Over time, the damage to customer trust and fines for legislative non-compliance alone can put an otherwise profitable company out of business.
How can you keep your organization safe when cyber fraud no longer leaves a signature? Security experts like GrackerAI's Gupta and James Scobey, CTO at B2B cybersecurity firm S2i2, think the answer lies in the same low-tech systems once considered too simple for large-scale business operations.
Live voice and video calls have remained the gold standard for identity verification throughout most of corporate history. These techniques have been central for authenticating high-value transactions, exchanging sensitive medical data, or discussing matters of legal importance.
Also: 43% of companies have already experienced AI-enabled cyberattacks
But in the last five years, that standard has been quietly eroded by deepfakes that have grown better at mimicking human behavior at an uncanny speed. There used to be certain tells, like background noises, synthetic voice modulation, and the distinct lack of breathing sounds. However, deepfake technology has outpaced these tried-and-tested tells, and it's difficult to tell when your digital coworker is real.