AI-Cloned CEOs Stole $129M From Finance Teams

The calls followed a pattern. An accounting employee would pick up the phone and hear what sounded exactly like their company's chief executive, speaking with authority and urgency about a wire transfer that needed to go out immediately. If the employee hesitated, a follow-up caller, this time impersonating a partner at a major law firm, would reinforce the instruction. The money moved. Then it was gone.

A federal indictment names three foreign nationals who ran this scheme for approximately 18 months starting in July 2024, targeting thousands of companies and stealing more than $129 million. Prosecutors allege the group used AI to clone the voices and likenesses of corporate CEOs and attorneys, then called finance and accounting staff with instructions to wire funds to accounts the defendants controlled.

The scam worked in two layers. First, the AI-generated voice of a known executive made the request sound legitimate and urgent. Second, an impersonated attorney gave the transaction a veneer of legal process, the kind that discourages questions. Neither caller was who they claimed to be, but the finance employees on the receiving end had no tool to test that claim in real time.

What the Finance Team Needed

This is the gap a codeword protocol fills. Trust Onion gives finance teams a standing challenge they can use before approving any high-stakes request: "What are the words?" The real CFO or CEO knows the current three rotating codewords. A cloned voice, however convincing, cannot know words it never heard, and those words change every 60 seconds. One question, asked before the wire goes out, is the check the employees in this scheme did not have.

The indictment names the defendants as one defendant, a second defendant, and a third defendant. The case shows that AI voice cloning is no longer a theoretical risk. It was the instrument of an active, large-scale fraud operation that ran for a year and a half before charges were filed.

The Attorney Layer

What made this scheme effective, according to reporting from Moneylife, was the second-caller tactic. After the fake CEO established the request, a fake attorney followed up to validate it. For a finance employee trained to treat executive and legal communications as authoritative, that combination was designed to suppress doubt. No callback policy or email confirmation step was described as part of the victims' standard process.

A callback to a number already on file is a reasonable first line of defense, but it does not prove the original caller was legitimate. Any verification step needs to happen before the wire is sent, not after. The question "What are the words?" makes that verification instant.

The $129 million figure is a threshold, not a ceiling. The scheme was still running when the indictment landed. Companies that had already sent the money had no quick path to recover it, if any path at all. Every organization targeted had a process that failed, and explaining that to a board or an auditor is its own problem.

A controlled voice-cloning safety demonstration

Try our tool to see just how easy voice cloning is. This controlled safety demonstration exists only to show how easy voice cloning is and to make the risk tangible. It is not a real incident or proof of anyone's identity. Your submitted sample is handled transiently to generate the fixed demonstration audio. Trust Onion does not identify you or the person from the sample, and does not keep the submitted recording or generated audio. The temporary voice clone is deleted from our voice provider as soon as the demo finishes; if that deletion fails, it is retried automatically until the provider confirms it.