Three men charged in US with using AI to impersonate CEOs and steal millions

AI-generated voice and video impersonation has moved from a theoretical security risk to an alleged nine-figure fraud operation. A newly unsealed US Department of Justice indictment charges three foreign nationals with using AI to mimic corporate CEOs and attorneys, persuading employees to authorize payments totaling more than $100 million. For builders deploying generative media or voice systems, the case is a clear reminder that realistic output can defeat approval processes built around familiar faces and voices. The defendants, Jacob Soall, Sharon Hazan and Guy Sheetrit, allegedly altered executives' voices and images while targeting multiple companies. Soall and Hazan were arrested in Spain over the summer, while Sheetrit was arrested in the UK in September. The charges point to a practical security response: organizations should treat calls and video meetings as untrusted channels, require out-of-band confirmation for large transfers, and use multi-person approvals rather than re