U.S. prosecutors indict three people, alleging they used AI to impersonate a CEO and a lawyer to defraud companies of more than $129 million
The U.S. Attorney’s Office for the Southern District of New York publicly unsealed an indictment on November 5, 2026, charging three foreign nationals, Jacob Soall, Sharon Hazan, and Guy Sheetrit, with participating in a transnational fraud scheme. Prosecutors said the three used artificial intelligence, described by prosecutors as “super intelligence,” to impersonate the voices or likenesses of corporate chief executive officers and lawyers at major international law firms, inducing company employees to transfer funds into accounts they controlled, and are suspected of defrauding more than $129 million. The three were arrested in Spain in June and August 2026 and in the United Kingdom in September 2026, respectively; extradition requests are pending.
According to the indictment, the scheme was carried out from around July 2024 to February 2026. The suspects typically first contacted employees in a company’s finance or accounting department, impersonating the CEO and claiming the company was advancing a highly confidential transaction and needed assistance; then, by phone or video call, they used AI to mimic the CEO’s voice and sometimes likeness. Under the false CEO identity, they directed employees to contact the “lawyer handling the transaction,” and a confederate then impersonated that lawyer to explain the details, stress confidentiality, and instruct that large sums of money be wired to designated accounts. Soall mainly impersonated a British lawyer and possessed a handwritten script titled “Lawyer Call,” which included content persuading victim employees not to disclose the matter to colleagues. After the funds were wired out, they were converted into cryptocurrency to hide their source and then distributed.
The indictment lists some of the victim losses, including about $4.82958 million from a hospital system in Missouri, about $2.68458 million from a bank in Rhode Island, about $2.968497 million from a hospital system in California, about $11.578474 million from a fintech company in the United Kingdom, about $4.639918 million from a mutual fund company in Massachusetts, and about $5.242747 million from a pet food company in Florida. The three face seven counts, including conspiracy to commit wire fraud, wire fraud, aggravated identity theft, conspiracy to launder money, and money laundering, with maximum penalties of decades in prison. Prosecutors emphasized that the charges are only allegations, and the defendants are presumed innocent until convicted.
Original source:
Title: Three Foreign Nationals Arrested And Charged With Scheme To Use Super Intelligence To Defraud Companies Of $129 Million Publisher: U.S. Attorney’s Office, Southern District of New York Date: 2026-11-05 URL: https://www.justice.gov/usao-sdny/pr/three-foreign-nationals-arrested-and-charged-scheme-use-super-intelligence-defraud Source nature: official indictment announcement Verification status: full text read in this instance, accessed on 2026-11-07 Supporting content: suspects’ identities and arrests, the method of using AI to impersonate CEOs and lawyers, the “Lawyer Call” script, victim loss amounts, money flow, and charges
ABAB AI Insight
The indictment in this case shows that the suspects built a complete chain of “CEO instructions + lawyer professional endorsement + confidentiality requirement for a secret transaction.” If employees followed normal internal procedures to “confirm upward,” they were facing multiple roles controlled by the same group rather than independent third parties. AI was used to mimic voices and likenesses, lowering the barrier to fabricating a single identity while also reducing the cost of coordinating multiple roles.
Compared with traditional business email fraud, this method introduced a real-time voice or video verification step and cited the identity of lawyers from a real major law firm as an authoritative source. The indictment says Soall’s handwritten script was specifically designed to deter employees from communicating with colleagues, directly targeting the enterprise’s internal cross-check mechanism. After the funds entered accounts in multiple countries, they were quickly converted into cryptocurrency, making tracing more difficult.
For targets such as medical institutions, banks, and fintech companies, “boss confirmation” or “lawyer involvement” in financial approval processes was once a security checkpoint for large transfers. When those checkpoints themselves can be forged from outside, the effectiveness of a single-confirmation rule declines. The prosecution data show that individual victim losses ranged from about $2.7 million to more than $11 million, indicating that the scheme targeted departments capable of handling large sums of money.
At present, the three have not been convicted, and the case remains at the charging stage. For the industry, the observable question is whether companies will upgrade “real-time voice/video confirmation” from default trust to independent multi-channel verification (for example, call-backs to known official numbers and secondary confirmation through internal systems), and what practical interception capability cryptocurrency tracing has in cross-border money laundering. The indicator to watch is whether, in subsequent similar cases, victim companies still rely on a single “boss or lawyer confirmation” as the final release condition.
ABAB News · Cognitive Laws
When the CEO and lawyer roles can both be forged by the same controlling party at the same time, the company’s internal cross-check mechanism fails.
AI lowers the cost of forging a chain of organizational authority, making confidentiality requirements themselves a tool to prevent external confirmation.
Once the approval checkpoint for large transfers is fully simulated from outside, the original rule of “upward confirmation” no longer constitutes an effective barrier.