# AI voice-cloning scam takes $129 million from companies; at CEO ONE, the technology works for the reader

A New York Times report shows how a criminal ring used fake voices and videos of executives to divert money from companies in the United States and Europe. The same technology, used with permission and transparency, is what narrates CEO ONE's articles.

- Section: Future
- Published: November 11, 2026
- Address: https://ceoone.com.br/en/article/ai-voice-cloning-scam-takes-129-million-from-companies-at-ceo-one-the
- Em português: https://ceoone.com.br/artigo/golpe-com-vozes-clonadas-por-ia-tira-us-129-milhoes-de-empresas-no-ceo-o.md

![A mobile phone on a dark wooden desk in a dimly lit office, its screen lit up showing a call and a sound wave, next to a pen and a leather folder](https://pub-b80e49b2737940b3ab5d3f20d0578192.r2.dev/artigos/golpe-com-vozes-clonadas-por-ia-tira-us-129-milhoes-de-empresas-no-ceo-o-foto.webp)

Produced by AI

## Key points

- A criminal ring used AI-cloned voices and videos of executives to divert more than $129 million from companies in the United States and Europe over about 18 months, according to an indictment reported by The New York Times.
- Three defendants were arrested in June and await extradition to the US; according to Dartmouth's Hany Farid, less than a minute of audio is now enough to clone a voice.
- At CEO ONE, the same technology is used with permission and transparency: the Listen to the article button plays an AI-generated narration, a project by Daniel Farias, the site's IT director.

The phone rang in the accounts department of a regional hospital system in Missouri, and the voice on the other end was familiar: it belonged to the organization's chief executive. He spoke of an urgent, confidential matter that required an immediate wire transfer. Shortly afterwards, a lawyer called to complete the transaction. By the time the real executives realized what had happened, $4.8 million had vanished. Neither man had made the call: both voices had been generated by artificial intelligence.

The case was reported by The New York Times on November 11, 2026, based on an indictment from the US Attorney's Office for the Southern District of New York. According to prosecutors, the same script was repeated for about 18 months, starting in July 2024, and cost companies in the United States and Europe more than $129 million. There were also attempts to draw employees into video calls with deepfakes of the chief executive.

## How the scam worked

The approach began with an email sent to a "victim employee", usually in accounting or financial management. The message asked her to sign a non-disclosure agreement and then call a lawyer. The defendants used the names of real people at real firms, often law firms ranked among the 50 largest in the world. One example cited in the indictment is an email sent in October 2024 to a major New York investment bank.

The victims, whose names were not disclosed, include a British fintech company ($11.6 million), a pet food manufacturer in Florida ($5 million), a hospital system in California ($3 million), a mutual fund manager in Massachusetts ($4.7 million) and a bank in Rhode Island ($2.7 million).

The money flowed to accounts controlled by accomplices in several countries, including Austria, China, Greece, Portugal, Singapore, Slovakia, Spain, the United Arab Emirates and the United States. According to prosecutors, it was then converted into cryptocurrency to conceal the source, ownership and control of the funds, and divided among the group.

Three men were arrested in June. Jacob Soall, 41, of Marbella, Spain, and Britain, and Sharon Hazan, 54, of Israel, were detained in Spain. Guy Shetrit, 42, of Israel and the United Arab Emirates, was arrested in Britain. All three are awaiting extradition to the United States, where they face charges of wire fraud, aggravated identity theft and money laundering. The newspaper was unable to reach lawyers for the defendants.

## Less than a minute of audio

Until a few years ago, imitating a voice required hours of high-quality recordings. That is why the targets were executives and lawyers with a high public profile, whose audio was available from earnings calls, talks and interviews. Hany Farid, a professor of computer science at Dartmouth and chief science officer at GetReal Security, a cybersecurity company specializing in AI and deepfake attacks, told the NYT that this barrier has fallen. "Now we are all potential targets," he said. "I would comfortably say with less than a minute, you can clone somebody."

Video has gone the same way. Instead of thousands of images of a person from different angles and in different lighting, a single photo is now enough to create a fake avatar, according to Farid. The software is easy to use, and a new laptop can handle the processing.

Trusting your ear does not solve the problem either. Many people believe they can tell the difference, but Farid's studies show otherwise. "People are just slightly above chance performance," he said. "So they're really, really bad at it."

## What the case teaches companies

The fraudsters' script always has the same ingredients: urgency, secrecy and a transfer outside the normal channels. That is precisely where the defense lies. A transfer request that arrives by phone or video, however familiar the voice may sound, should be confirmed through another channel, by calling back a number the company already knows. Transfers above a set amount should require approval from two people, and no non-disclosure agreement exempts anyone from that procedure. It is also worth agreeing with the board that no executive will ever request an urgent payment outside the process, so that the finance team knows it can, and should, be suspicious.

## The same technology, working for the reader

The artificial intelligence that imitates voices to deceive can also serve those who read. At CEO ONE, articles can be listened to: on every story on the site, the Listen to the article button plays a natural-sounding narration generated by artificial intelligence.

The project was developed by Daniel Farias, CEO ONE's IT director. In the first phase, using just two podcasts as a basis, the technology learned to accurately reproduce the voice of Ricardo Battistini, CEO of the Vip Battistini ecosystem, who authorized its use, and it was this voice that narrated the site's first articles. Today the narrations have a new voice, chosen to further improve the listening experience.

The difference between the two uses lies in consent and transparency: the voice was used with the speaker's permission, and the reader knows the narration is produced by AI. The purpose is simple: to let everyone stay informed while driving, walking or resting. To try it, just open any article at ceoone.com.br and tap Listen to the article.

## Sources and editorial notes

- Reference reporting: "A.I. Helped Steal More Than $129 Million From Companies", by Adeel Hassan, The New York Times, November 11, 2026, based on the indictment from the US Attorney's Office for the Southern District of New York.
- Analysis and the narration case: content developed by CEO ONE.

CEO ONE · https://ceoone.com.br