AI voice cloning deepfakes drive rise in sophisticated cyber scams

Cybercriminals are increasingly using artificial intelligence to clone voices, a technique known as audio deepfakes, to conduct sophisticated scams and virtual kidnappings. Security firm ESET Latinoamérica warns that the low cost and high accessibility of these tools have turned these incidents from isolated cases into a growing regional threat.

Attackers can create convincing replicas using only a few seconds of audio extracted from public social media videos or professional content. To increase the effectiveness of the deception, criminals often use emotional manipulation, such as mixing the cloned voice with background noises like crying or screaming, and utilizing open-source intelligence (OSINT) to tailor their stories to specific victims.

In Spain, the National Cybersecurity Institute (INCIBE) reports that over 135 million fraudulent calls have been blocked so far this year, with approximately one-third utilizing this technology. A global study cited by ESET indicates that in the United States, one in four people reported receiving a call with a cloned voice in the last year, and 24% of respondents could not distinguish a synthetic voice from a real one.

To mitigate these risks, experts recommend establishing family security protocols, such as a pre-agreed safe word, to validate identities during suspicious calls.