On November 6, 2026, the U.S. Attorney for the Southern District of New York unsealed an indictment charging three people with wire fraud, aggravated identity theft, and money laundering for stealing approximately $129 million over 18 months using AI-generated voice and video impersonation of corporate executives and attorneys.
How the scheme worked
The three defendants, Jacob Soall, Sharon Hazan, and Guy Sheetrit, targeted the accounting departments of thousands of companies worldwide. The initial contact was an email purporting to come from the target company's CEO, requesting help with a confidential financial transaction. When the employee engaged, the defendants coordinated a phone or video call where AI cloned the CEO's voice and sometimes mimicked the CEO's image. The employee was then referred to an attorney at a prominent law firm, whom the group was simultaneously ready to impersonate. Soall specialized in impersonating U.K. attorneys and reportedly used a handwritten script titled 'Lawyer Call' during calls. The impersonated attorney directed the employee to wire funds to accounts controlled by the group. The funds were then converted to cryptocurrency to obscure the trail.
What made it scalable
The scheme's reach, thousands of targeted companies across multiple countries, came from the operational structure. Voice cloning and video deepfakes lowered the credibility barrier that previously made executive impersonation scams easier to detect. The two-stage impersonation pattern, CEO email followed by attorney call, created a chain of apparent authority that bypassed the intuition that typically flags a single unsolicited request. Victims included hospital systems in Missouri and California, a bank in Rhode Island, a fintech company in the United Kingdom, a mutual fund company in Massachusetts, and a consumer business in Florida. The diversity of targets indicates the group applied the same playbook across industries without meaningful customization per victim.
What defenders can do
- Establish an out-of-band verification protocol for wire transfers above a defined threshold. Calls that arrive through the same channel as the original request, including AI-generated video, should not be treated as independent confirmation.
- Train finance and accounting staff to recognize the two-stage pattern: an executive email followed by referral to an attorney or compliance contact for a wire instruction is a known attack template, not a standard business process.
- Register your executives' voice and image with your fraud team as reference samples. AI detection tools are improving but are not reliable enough to serve as a sole control. Policy and verification process are more durable defenses than technology detection alone.
The FBI noted this case illustrates how advanced technology can be used to commit large-scale fraud. The indictment does not change the fundamental attack model; it confirms it at a scale that makes the threat model concrete for defenders who have been treating AI impersonation as a future concern.