Purported AI Impersonations of Executives Allegedly Helped Defraud Companies Worldwide of About $129 Million

Federal prosecutors in New York alleged that a fraud ring used AI to mimic executives' voices and sometimes their images on calls with company employees. Accomplices posing as lawyers then directed employees to wire money for sham transactions. The indictment attributes about $129 million in losses worldwide to the scheme, including about $4.83 million from a Missouri hospital system.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The perpetrators used AI voice and image cloning to impersonate specific CEOs and attorneys to execute targeted financial fraud against corporate employees, stealing over 129 million dollars.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was caused by human actors who intentionally deployed AI voice and image cloning tools to execute a massive financial fraud scheme.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes the use of AI to generate deepfakes (cloned voices and images), which falls under the transparency obligations of Limited Risk systems.

AI system and alleged parties

• AI system: None named
• AI purpose: Voice Generation; Deepfake Video Generation
• Behaviour type: Tool
• Alleged developer: Voice cloning technology developers, Synthetic media generation technology developers, Synthetic audio generation technology developers
• Alleged deployer: Voice cloning impersonation scammers, Synthetic media creators, Sharon Hazan, Scammers, Jacob Soall, Impersonation scammers, Guy Sheetrit, Deepfake creators, Cybercriminals, Alleged $129 million executive impersonation fraud ring
• Alleged harmed parties: Victims of impersonation scams, Victims of deepfake-enabled fraud, Privacy, Organizations, Healthcare organizations, Financial institutions, Epistemic integrity, Companies

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Severe. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Severe, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Minor, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: The reports explicitly describe total financial losses of at least 129 million dollars from the overall scheme across thousands of targeted companies.

Directly caused: The scheme directly caused at least 129 million dollars in financial losses to the victim companies, averaging millions of dollars per successful fraud occurrence.

Inferred additional harm: Additional unquantified financial losses likely include legal fees, forensic investigation costs, and security upgrades for the thousands of targeted companies.

Privacy

Reported: The reports explicitly describe aggravated identity theft, where the defendants used the identifying information of real CEOs and attorneys.

Directly caused: The defendants used the names, voices, and likenesses of real corporate executives and attorneys without authorization to conduct the fraud.

Epistemic

Reported: The reports describe the fabrication of voices and images of corporate executives to mislead employees into believing sham transactions were real.

Directly caused: The defendants used AI to mimic the voices and images of CEOs in phone and video calls to convince employees that the transactions were legitimate.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1000
• People reportedly exposed: 5000

Potential causes

Management

• Inadequate Risk Assessment: Management did not prepare for AI-driven impersonation of key executives.
• Weak Back-Office Controls: Failure of back-office verification in victim companies except for one.

Technology

• AI Voice Mimicking: AI was used to mimic the voice of the Impersonated CEO on calls.
• AI Image Alteration: AI was used to alter and mimic the image of the Impersonated CEO.

Data Inputs

• Impersonated Identity Details: Using real names of European attorneys and CEOs to establish trust.
• E-Signature Emails: Purported CEO emails sent via DocuSign to initiate contact.

Human Factors

• Employee Trust in AI Media: Employees believed the realistic AI-cloned voices and images on calls.
• Adherence to Secrecy Requests: Employees kept transactions secret as directed by the impersonated attorney.

Process and Methods

• Inadequate Verification Process: Lack of verification allowed wires to go to external accounts.
• Failure to Flag Spoofed Emails: E-signature emails from fake CEOs successfully reached target employees.

Regulatory Environment

• Weaponization of Emerging Tech: Criminals weaponized emerging AI technologies to bypass standard security.
• International Money Laundering: Conspirators used global bank accounts and crypto to launder stolen funds.

Information quality

• Classification confidence: High
• Reason for confidence: The assessment is based on a detailed federal indictment and reputable news reports, which provide consistent and specific facts about the fraud scheme, the victims, and the role of AI.
• Ambiguities identified: The specific AI models or software platforms used to generate the voices and images are not identified in the reports.

A highly sophisticated global fraud scheme utilized AI voice and video cloning to impersonate corporate executives, stealing over $129 million from thousands of companies. While primarily a criminal financial enterprise rather than a state-sponsored threat, the scale and success of the operation demonstrate the potent economic threat posed by advanced generative AI tools in the hands of malicious actors.

• Overall national security impact: Substantial
• Scope: Multiple nations
• Primary target: United States
• Other affected: United Kingdom, Spain, Portugal, Singapore, Greece, United Arab Emirates, Austria, Slovakia, China
• Alleged perpetrator: Guy Sheetrit and co-conspirators

Threat characteristics

• Imminence: Long-term. The perpetrators have been arrested, representing a resolved immediate threat but highlighting an ongoing strategic capability concern for deepfake-enabled fraud.
• Autonomy: Human-controlled. AI voice and video cloning tools were directly operated by human conspirators to generate specific deceptive content.
• Novelty: Evolved capability. Represents a highly sophisticated and scaled evolution of existing business email compromise and voice-phishing techniques using advanced generative AI.

Impact by dimension

• Physical security: Minor. While hospital systems in Missouri and California were targeted, the impact was limited to financial fraud rather than operational disruption of critical healthcare infrastructure.
• Information security: Minor. Utilized sophisticated AI voice and video cloning to impersonate executives. While highly deceptive, it was deployed for financial fraud rather than state-sponsored intelligence or disinformation operations.
• Sovereignty: Negligible. No core government functions, electoral systems, or sovereign decision-making processes were targeted or compromised in this incident.
• Economic security: Substantial. The scheme successfully defrauded thousands of companies of over $129 million, targeting key sectors including finance, healthcare, and technology, demonstrating a highly scalable threat to business security.
• Societal stability: Negligible. The incident did not involve mass surveillance, systematic discrimination, or threat to civil liberties, and had negligible direct impact on broader societal stability.