This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Since April 2025, malicious actors have been conducting a sophisticated vishing and smishing campaign targeting senior U.S. government officials. The attackers use AI-generated voice cloning to impersonate officials, aiming to gain access to personal and official accounts, including a reported breach involving White House Chief of Staff Susie Wiles. The FBI has issued a public alert regarding this ongoing threat, noting that the campaign leverages social engineering to compromise sensitive communications and potentially elicit funds or information.
According to a May 15, 2025 public alert from the FBI and IC3, malicious actors have been using purported AI-generated voice messages and smishing tactics since April 2025 to impersonate senior U.S. officials. The campaign reportedly targets government personnel and their contacts to gain access to personal or official accounts. The FBI warns the scheme may involve credential theft and social engineering, with AI voice cloning used to increase believability.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Also classified as
The attackers successfully accessed Susie Wiles' personal contacts and targeted personal accounts of officials to obtain sensitive, non-public information.
The incident involves malicious actors using AI-generated voice cloning to impersonate specific senior U.S. officials to gain unauthorized access to personal accounts, which represents targeted manipulation and impersonation fraud.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident is driven by malicious human actors intentionally using deployed AI voice-cloning technology to conduct targeted phishing and impersonation campaigns.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
A sophisticated cyber-espionage and vishing campaign targeted senior U.S. officials, successfully breaching the contacts of White House Chief of Staff Susie Wiles. Utilizing AI-generated voice cloning, the attackers impersonated Wiles to target senators, governors, and executives. This represents a severe threat to government communications and intelligence security.
The reports describe malicious actors using AI-generated audio and voice-cloning techniques to mimic the voices of senior officials. The AI acts as a tool, performing the specific function of generating realistic voice outputs based on direct inputs provided by the human operators.
Limited Risk: The report describes the use of 'AI-generated content such as deepfakes' and 'AI-generated voice messages' to impersonate individuals, which falls under transparency obligations for synthetic content.
The AI system was used to generate realistic, cloned voices of senior U.S. officials to deceive targets during phone calls and voice messages.
The reports from CyberScoop, the FBI (IC3), and the Washington Post provide consistent, detailed accounts of the vishing/smishing campaign, including the specific breach of Susie Wiles' contacts. The role of AI voice cloning is clearly stated, and the targets and methods are well-documented.
None. The reports clearly point to a coordinated social engineering and cyber-espionage/phishing campaign using AI voice cloning.
Technology
AI Voice Cloning Technology
AI tools generate highly realistic voice clones to mimic senior officials.
Scalable Content Generation
Generative AI enables adversaries to fabricate fake audio and text at scale.
Use of Insecure Apps
Use of insecure third-party versions of encrypted messaging applications.
Data Inputs
Compromised Contacts Lists
Breaching personal cellphones provides actors with trusted contact details.
Publicly Available Photos
Actors use public photos in text messages to deceive targets and gain trust.
Human Factors
Cognitive Trust Exploitation
Targets trust messages that appear to come from known, high-profile contacts.
Inadequate Security Vigilance
Users fail to independently verify the identity of callers or senders.
Mixing Personal and Work Devices
Officials using personal phones to discuss sensitive government policies.
Process and Methods
Insecure Communication Habits
Conducting official government business on unapproved personal channels.
Lack of Out-of-Band Verification
Absence of strict protocols to verify sender identity via trusted channels.
Clicking on Unverified Links
Users clicking links in messages to transition to secondary platforms.
Management
Weak Policy Enforcement
Failure to enforce strict secure communication protocols among senior staff.
Inadequate Risk Mitigation
Lack of preparation for AI-facilitated social engineering threats.
Regulatory Environment
Gaps in AI Impersonation Laws
Current legal frameworks lack sufficient protections against AI impersonation.
Insufficient Deterrence
Existing penalties fail to prevent actors from deploying deepfake campaigns.
AI system involved
An AI system was involved, but the reports do not identify it.
Identified by automated extraction; can differ from the alleged developers above.
