Hospitals AI media incident — Nov 2026

Federal prosecutors in Manhattan have charged three men with orchestrating a sophisticated fraud scheme that utilized artificial intelligence to impersonate executives and attorneys, resulting in the theft of approximately $129 million from organizations worldwide. The conspirators allegedly targeted hospitals, banks, and other companies by leveraging voice and face cloning technology to deceive accounting staff into authorizing fraudulent wire transfers.

What happened

The operation, which ran from at least July 2024 to February 2026, relied on a consistent social engineering playbook. The process typically began with an email sent via an electronic signature service, appearing to originate from the target company's chief executive. This email would request assistance with a confidential transaction and include a nondisclosure agreement to discourage internal verification.

Following the initial email, the conspirators initiated phone calls where they used AI to mimic the voices and, in some instances, the faces of the executives. The fake CEO would then introduce the employee to a fake lawyer, often assuming the identity of a real attorney from a top-tier law firm. A handwritten script titled “Lawyer Call” was used to pressure employees, citing confidentiality and insider-trading rules to prevent them from consulting with colleagues.

Victims included a Missouri hospital system, a Rhode Island bank, and a British financial technology company, among others. Once the funds were wired to accounts across multiple countries, including Spain, Singapore, and the UAE, the money was converted into cryptocurrency. The indictment charges Jacob Soall, Sharon Hazan, and Guy Sheetrit with seven counts each, including conspiracy to commit wire fraud and money laundering. While the scheme was highly effective, it was not infallible; one New York investment bank successfully blocked a $2.85 million transfer when a back-office employee performed a manual review of the transaction. Prosecutors emphasized that the defense that worked was not a deepfake detector, but rather the intervention of a second pair of human eyes.