Study guide

9.4 Deepfakes, Generative AI Threats, and Scam Typologies

ACAMS Certified AML FinTech Compliance Associate CAFCA

Track your progress as you read

Finished sections get a check mark.

1 Chapter 1: CAFCA Exam Overview and Study Strategy

1.1 CAFCA Exam Format, Official Domains, and Study Strategy

2 Chapter 2: Domain 1: FinTech Business Models and Their Vulnerabilities

2.1 Neobanks, Challenger Banks, and Digital Banking Models 2.2 Payment Processors, Gateways, and Merchant Acquirers 2.3 Buy Now, Pay Later (BNPL) and Alternative Credit 2.4 BaaS Architecture, Embedded Finance, and Program Structures

3 Chapter 3: Domain 1: Financial Crime Types and the Regulatory Perimeter

3.1 Types of Financial Crime and Predicate Offences 3.2 Licensing, Registration, Charters, and Regulatory Sandboxes 3.3 Virtual Asset Service Provider (VASP) Framework & FinCEN MSB Rules

4 Chapter 4: Domain 1: Global AML Frameworks and Sources of Guidance

4.1 U.S. Regulatory Architecture: BSA, PATRIOT Act, and AMLA 2020 4.2 European Union Regulatory Framework: AMLD6, the AML Package, and MiCA 4.3 International Standards: UK FCA, Singapore MAS, and Hong Kong HKMA 4.4 Sources of Guidance for Building Anti-Financial Crime Processes

5 Chapter 5: Domain 1: Risk Management Framework, Risk-Based Approach, and Data Protection

5.1 Risk Management Framework: Lines of Defence, Policies, Assurance, and the MLRO 5.2 Risk-Based Approach, Risk Assessment, Risk Appetite, and Types of Risk 5.3 Handling Sensitive Information: PII, SPII, Privacy Laws, and Breach Reporting

6 Chapter 6: Domain 1: Bank Partners and Internal Threats

6.1 Regulatory Expectations and Enforcement on Sponsor Banks 6.2 How Banks Risk-Rate FinTechs: Onboarding, Ongoing Oversight, and De-Risking 6.3 Controlling Internal Threats: Insider Fraud, Collusion, and Data Misuse

7 Chapter 7: Domain 2: CIP, KYC, CDD, and EDD

7.1 Customer Identification Program (CIP) Under USA PATRIOT Act §326 7.2 The FinCEN CDD Rule & Beneficial Ownership Standards 7.3 KYC to EDD: Applying CDD and Enhanced Due Diligence Across Customer Types

8 Chapter 8: Domain 2: Identity Verification, Data Sources, and Device Data

8.1 Digital Onboarding & Electronic KYC (eKYC) Foundations 8.2 Biometric Face Matching and Liveness Detection 8.3 Verification Data Sources and How to Judge Their Reliability 8.4 Device Fingerprinting, Geolocation, and Behavioral Biometrics

9 Chapter 9: Domain 2: Onboarding Screening, Risk Ratings, and Fraud Red Flags

9.1 Screening at Onboarding: Sanctions Lists, PEPs, and Known Fraudsters 9.2 Customer Risk Ratings: Factors, Data Sources, and How Scoring Models Work 9.3 Detecting Synthetic Identity Fraud and Ineligible Applicants 9.4 Deepfakes, Generative AI Threats, and Scam Typologies

10 Chapter 10: Domain 3: Payment Screening

10.1 OFAC Sanctions Regimes & the 50 Percent Rule 10.2 Real-Time vs. Batch Screening Architecture in High-Velocity Rails 10.3 Screening Logic: Fuzzy Matching, Phonetics, and PEP Controls

11 Chapter 11: Domain 3: Transaction Monitoring Systems and Red Flags

11.1 Hybrid Transaction Monitoring: Rules Engines vs. Machine Learning 11.2 FinTech Typologies: Mule Networks, Funnel Accounts, and Rapid Pass-Through 11.3 Fraud-AML Convergence: BIN Attacks, ATO, and Credential Stuffing 11.4 Model Risk Management Under SR 11-7 & System Optimization

12 Chapter 12: Domain 3: Payment Methods and Transaction Risk Characteristics

12.1 Common Payment Methods, Crypto, and High-Risk Transactions Such as Aggregation 12.2 Transaction Risk Characteristics: Reversibility, Convertibility, Fund Integrity, and Privacy Coins 12.3 Wallet Types, Custody Architectures, and Counterparty Risks 12.4 Stablecoins, On/Off Ramps, and Asset-Referenced Tokens

13 Chapter 13: Domain 3: Monitoring Crypto Transactions

13.1 Blockchain Analytics & Transaction Tracing 13.2 The FATF Travel Rule for Virtual Assets 13.3 DeFi, Mixers, and Smart Contract Sanctions

14 Chapter 14: Domain 3: Alert Investigation, Audit Trails, and Monitoring MI

14.1 The Alert Investigation Process: From Unusual to Suspicious and When to Escalate 14.2 Building an Audit Trail for Suspicious Activity Reviews 14.3 Communicating Monitoring Results: KPIs, KRIs, OKRs, and Management Information

15 Chapter 15: Domain 4: Investigations and Customer Communication

15.1 Talking to Customers During Investigations and Offboarding: Financial Crime Risk vs Policy Breach 15.2 Reviewing KYC, Transactions, Open Sources, and Documents to Build an Investigation 15.3 Analytical Principles in Investigations: Avoiding Confirmation Bias and Other Traps

16 Chapter 16: Domain 4: SARs and STRs

16.1 SARs and STRs: What They Are, When and Where to File, and Why They Matter 16.2 Writing Effective SAR and STR Narratives

17 Chapter 17: Domain 4: Law Enforcement, Regulators, and Banking Partners

17.1 Operational AML Governance and Third-Party Risk Management 17.2 Working With Law Enforcement, Banking Partners, Regulators, and Other Third Parties 17.3 Types of Law Enforcement Requests: Subpoenas, Court Orders, 314(a), and International Requests 17.4 How Banking Partners' Terms of Service Relate to the FinTech's Own Terms

18 Chapter 18: Domain 5: Scaling Anti-Financial Crime Strategies

18.1 Assessing New Products and Features Before Launch 18.2 Refreshing the Financial Crime Risk Assessment as the Business Scales 18.3 Changes That Should Trigger a Financial Crime Risk Assessment Review 18.4 Record Retention Rules and Secure Data Storage 18.5 Outsourcing Controls: Surge Capacity, RegTech, and Independent Testing

9.4 Deepfakes, Generative AI Threats, and Scam Typologies

Key Takeaways Explain This Topic AI

Generative AI enables the creation of sophisticated synthetic identities by combining unassigned or deceased Social Security Numbers with LLM-generated biographical histories and forged utility documents that bypass legacy optical character recognition (OCR) verification.

Deepfake injection attacks bypass biometric selfie onboarding by utilizing virtual camera software and emulators to feed GAN-generated faces and real-time digital puppetry streams directly into application video capture APIs, circumventing physical presentation attack detection.

Romance baiting and 'pig butchering' (Sha Zhu Pan) scams operate through industrial-scale overseas fraud compounds, utilizing multi-month psychological grooming to funnel victim funds into fraudulent Web3 trading platforms displaying manipulated yields.

Laundered proceeds from pig butchering schemes are obfuscated through rapid blockchain peeling chains, unhosted intermediate wallets, nested Virtual Asset Service Providers (VASPs), and high-volume Southeast Asian Over-The-Counter (OTC) cash brokers.

FinCEN's pig-butchering alert (FIN-2023-Alert005) asks filers to use the key term FIN-2023-PIGBUTCHERING, and its deepfake alert (FIN-2024-Alert004) asks for FIN-2024-DEEPFAKEFRAUD, in SAR field 2 and the narrative.

Last updated: October 2026

The exponential acceleration of artificial intelligence, particularly deep learning models, Generative Adversarial Networks ( GANs ), and Large Language Models ( LLMs ), has fundamentally altered the threat landscape for financial technology compliance. While machine learning strengthens transaction surveillance, threat actors exploit these same technologies to bypass remote identity verification, automate social engineering, and execute industrial-scale investment frauds. Anti-financial crime professionals must understand the technical architecture of generative AI threats and the money laundering mechanics underpinning transnational fraud typologies such as romance baiting and investment scams.

Generative AI and Synthetic Identity Fabrication

Synthetic Identity Fraud (SIF) has evolved from basic manual document alteration into an automated, AI-driven discipline. Rather than stealing an entire living person's identity, modern fraud syndicates synthesize entirely fictional personas.

The Evolution from Identity Theft to Identity Synthesis

Compounding Real and Synthetic Data : Threat actors obtain legitimate, dormant Social Security Numbers ( SSNs )—frequently belonging to children, deceased individuals, or unissued SSN ranges (such as those randomized by the Social Security Administration post-June 2011)—and pair them with fabricated names, dates of birth, and addresses.

LLM-Generated Biographical Footprints : Modern syndicates deploy specialized LLM agents to automatically generate comprehensive, internally consistent biographical profiles. The AI establishes plausible digital footprints: creating social media accounts, registering personal websites, generating professional résumés on employment platforms, and producing realistic behavioral histories.

AI-Forged Supporting Documentation : Using image-generation diffusion models, syndicates fabricate synthetic supporting documentation required during electronic Customer Due Diligence ( eCDD ). These include hyper-realistic utility bills, vehicle registrations, bank account statements, and pay stubs. Unlike crude manual photo edits, AI-generated documents contain perfectly rendered typography, valid barcode metadata, consistent regional sales tax calculations, and authentic micro-textures that easily evade traditional Optical Character Recognition ( OCR ) tampering rules.

Tradeline Leasing, Credit Piggybacking, and Bust-Outs

Once a synthetic identity is created, the syndicate must establish a credit history:

Authorized User Tradelines ("Piggybacking") : The synthetic identity is added as an authorized user to a high-limit, seasoned credit card owned by a third-party broker for a fee. The primary cardholder's positive credit history transfers to the synthetic identity's credit file.

Credit Incubation : Over a 12- to 24-month incubation period, the synthetic persona establishes credit accounts with low-risk digital lenders and FinTech credit builder programs, diligently paying balances to cultivate prime credit scores (often exceeding 720 FICO).

Coordinated Bust-Out : Once credit limits reach substantial amounts, the syndicate simultaneously draws down maximum personal loans, credit card cash advances, and lines of credit across dozens of FinTech lenders. The funds are channeled into payment apps and crypto on-ramps, leaving lenders with non-collectible debt attached to a debtor who does not legally exist.

Biometric Bypasses: Deepfake Injection Attacks vs. Presentation Attacks

As financial technology firms migrated to mobile, branchless onboarding, the security of the Customer Identification Program ( CIP ) under USA PATRIOT Act Section 326 came to depend heavily on remote facial biometrics: capturing a photograph of a government ID document paired with a live "selfie" video.

Presentation Attacks vs. Digital Injection Attacks

Compliance officers must distinguish between two fundamentally different biometric spoofing vectors:

Presentation Attacks (Physical Spoofs) : An attacker presents an analog, physical spoof artifact directly in front of an authentic physical camera lens. Examples include holding up a printed color photograph, playing a pre-recorded video on an iPad screen facing the camera, or wearing a 3D sculpted silicone mask. Presentation Attack Detection ( PAD ) systems counter these by analyzing depth, paper texture, screen refresh rates, and ambient light reflections.

Digital Video Injection Attacks : A far more dangerous vector that completely bypasses the physical camera lens and hardware sensors. Threat actors utilize software emulators, virtual camera drivers (such as OBS Virtual Camera), or rooted mobile operating environments to feed a pre-rendered or real-time deepfake video stream directly into the operating system's video capture API (such as WebRTC in mobile browsers). Because no physical camera lens is used, standard physical PAD defenses fail completely.

Deepfake Technologies: GANs, Face Swapping, and Digital Puppetry

Modern injection attacks utilize advanced machine learning architectures:

Generative Adversarial Networks (GANs & StyleGAN) : Algorithms that generate photorealistic, ultra-high-resolution synthetic human faces from scratch, complete with realistic skin pores, hair strands, and lighting reflections.

Real-Time Digital Puppetry and Face-Swapping : Software (such as DeepFaceLive) that captures an attacker's live webcam feed, tracks their facial landmark movements in real time, and projects the target identity's face onto the attacker. When the onboarding system prompts the user to "smile", "blink", or "turn your head left", the attacker performs the motion, and the deepfake model animates the synthesized face synchronously.

Defensive Countermeasures: PAD, Active Liveness, and Hardware Attestation

To defend against deepfake injection attacks, FinTech compliance architectures deploy multi-layered verification:

Cryptographic Hardware Attestation : Utilizing device-level APIs—such as Apple App Attest / DeviceCheck on iOS and Google Play Integrity API on Android. These frameworks use hardware security enclaves to cryptographically sign video capture data, proving to the onboarding server that the video feed originated from an authentic, physical camera sensor running on an unrooted, non-emulated operating system.

Dynamic Challenge-Response Liveness : Projecting randomized, shifting light patterns or specific color sequences onto the applicant's face from the smartphone display, analyzing whether the applicant's skin and corneas reflect the expected color sequence (reflectance analysis), sometimes combined with remote photoplethysmography ( rPPG ), which looks for pulse-related color changes in skin.

Frame-Level Compression Analysis : Detecting spatial and temporal artifact anomalies, edge jitter, and inconsistent frame rates characteristic of GAN synthesis and virtual camera rendering pipelines.

Voice Cloning and Social Engineering in Customer Servicing

While visual deepfakes target onboarding, generative audio synthesis and voice cloning target telephone customer support channels and Interactive Voice Response ( IVR ) systems.

Acoustic Synthesis and Zero-Shot Voice Cloning

Modern deep learning diffusion models and neural audio codecs can clone a human voice with staggering fidelity using as little as 3 to 10 seconds of reference audio . Threat actors harvest clean voice samples of corporate executives or high-net-worth individuals from public podcasts, earnings calls, YouTube videos, or social media clips.

Targeting Call Centers and Bypassing Voice Biometrics

Helpdesk Social Engineering : Attackers call FinTech support desks utilizing real-time text-to-speech voice changers, perfectly impersonating the account holder. Presenting stolen PII, they socially engineer customer service representatives into resetting registered phone numbers, altering email addresses, or approving manual wire transfer exceptions.

Defeating Voice Biometrics : Modern banking platforms that deploy automated voiceprint authentication (measuring pitch, formant frequencies, and vocal tract dimensions) are vulnerable to neural voice clones that match the acoustic spectral density of the genuine customer, allowing unauthorized access to telephone banking menus.

The "Pig Butchering" (Sha Zhu Pan) Typology: Anatomy of Industrialized Scams

Originating in Southeast Asia and designated in Mandarin as Sha Zhu Pan (杀猪盘, literally "pig butchering"), this financial crime typology has evolved into an industrialized, multi-billion-dollar global scam apparatus.

Industrialized Criminal Compounds in Southeast Asia

Unlike opportunistic individual scammers, pig butchering operations are run by transnational organized crime syndicates operating from heavily fortified, casino-style compounds located in special economic zones in Southeast Asia (notably in Myanmar, Cambodia, and Laos ). Many front-line operators are themselves victims of human trafficking, coerced under threat of violence to execute sophisticated social engineering scripts.

  • The Five-Stage Lifecycle of Romance Investment Scams THE PIG BUTCHERING (SHA ZHU PAN) LIFECYCLE
  • │
  • ┌──────────────────┬─────────────────────┼─────────────────────┬──────────────────┐
  • ▼ ▼ ▼ ▼ ▼
  • 1. Target Find 2. Grooming 3. The First Taste 4. Slaughter 5. Extortion
  • ┌───────────────┐ ┌──────────────────┐ ┌──────────────────┐ ┌───────────────┐ ┌──────────────────┐
  • │ • Unsolicited │ │ • Build intimacy │ │ • Small deposit │ │ • Liquidate │ │ • Freeze account │
  • │ WhatsApp / │ │ • Flaunt wealth │ │ • Fake DApp yield│ │ 401(k), home│ │ • Demand 30% fake│
  • │ SMS / Dating│ │ • Casual mention │ │ • Successful test│ │ equity, loan│ │ AML/tax fee │
  • │ • AI profiles │ │ of crypto gains│ │ withdrawal │ │ • Large wires │ │ • Total cut-off │
  • └───────────────┘ └──────────────────┘ └──────────────────┘ └───────────────┘ └──────────────────┘

Target Acquisition ("Finding the Pig") : Scammers initiate contact through unsolicited text messages ("wrong number" pretexts sent via SMS, WhatsApp, or Telegram), professional platforms (LinkedIn), or dating applications (Tinder, Bumble, Hinge). Attackers utilize glamorous, AI-generated profile photos portraying high-net-worth lifestyles.

Grooming and Psychological Manipulation ("Fattening the Pig") : Over weeks or months, the scammer builds intense psychological rapport, feigning romantic interest or close friendship. The scammer subtly displays enormous wealth (luxury vehicles, exotic travel), attributing their financial independence to an exclusive, proprietary cryptocurrency trading platform or decentralized finance ( DeFi ) liquidity pool.

The First Taste (Fabricated Yields) : The scammer patiently guides the victim to open an account at a regulated domestic crypto exchange (e.g., Coinbase, Kraken), purchase fiat-pegged stablecoins (USDT or USDC), and transfer the assets to an external platform. This destination is typically a fraudulent Web3 DApp accessed via mobile crypto wallet browsers, or a cloned exchange interface hosted on a lookalike domain. The fraudulent platform manipulates internal account balances to show astronomical daily returns (15% to 30%). To eliminate suspicion, the scammer instructs the victim to execute a real withdrawal of their initial $1,000 principal plus profit back to their bank account.

The Slaughter ("Killing the Pig") : Convinced of the investment's validity, the victim liquidates life savings, retirement 401(k) accounts, and home equity lines of credit ( HELOC ), wiring hundreds of thousands or millions of dollars into the scam platform.

The Extortionate Freeze : When the victim attempts to withdraw substantial funds, the platform freezes the account, claiming the funds are locked by "tax authorities", "international AML regulators", or "smart contract audit thresholds." The platform demands escalating upfront cash payments (e.g., an additional "30% capital gains tax deposit") to unlock the balance. Once the victim is drained of all financial assets, the scammers sever all communication and delete the fraudulent platform.

Blockchain Tracing and Money Laundering Mechanics

Once victim funds are converted into cryptocurrency, syndicates execute complex on-chain layering techniques to obscure origins:

Peeling Chains and Unhosted Wallet Hops

Upon receiving stablecoins from the victim, the syndicate's automated deposit wallet initiates a peeling chain : an automated transaction sequence where a small fraction of the funds is peeled off to pay operational gas fees or subordinate wallets, while the vast majority is swept to intermediate unhosted wallets across hundreds of rapid hops within minutes.

Nested VASPs and Unlicensed OTC Desks

To convert virtual assets back into fiat currency without triggering standard AML controls, syndicates utilize nested Virtual Asset Service Providers (VASPs) . A nested VASP is an entity that operates master accounts or corporate omnibus accounts on large, regulated international crypto exchanges, providing sub-account trading services to third parties without conducting independent CDD or KYC. Stolen stablecoins are funneled through nested Over-The-Counter ( OTC ) trading desks operating in Southeast Asia.

Cross-Chain Bridges and Fiat Off-Ramping

Proceeds are fragmented across multiple blockchain networks (e.g., swapping TRC-20 USDT on TRON for ERC-20 tokens on Ethereum or Bitcoin) using decentralized cross-chain bridges. Ultimately, the funds are liquidated into physical cash or local currencies through underground banking networks, trade-based money laundering schemes, or casino junket operators in Macau and Cambodia.

Emergency Compliance Interventions and Regulatory Filings

When a financial technology platform detects transactions or customer interactions indicative of pig butchering, immediate intervention is required to safeguard the victim and fulfill statutory reporting duties.

FinCEN Alerts on Pig Butchering and Deepfakes

In September 2023, FinCEN issued FIN-2023-Alert005 , an alert on the virtual currency investment scam known as pig butchering, with behavioral, technical, and transactional red flags. In November 2024, it issued FIN-2024-Alert004 on fraud schemes using deepfake media, including altered or synthetic identity documents and deepfake video used to defeat identity checks. Older and isolated customers are frequent targets, so FinCEN's guidance on elder financial exploitation is also relevant.

Key reporting points:

Key Terms : Enter FIN-2023-PIGBUTCHERING for pig-butchering activity, or FIN-2024-DEEPFAKEFRAUD for deepfake-related fraud, in SAR field 2 (Filing Institution Note to FinCEN) and in the narrative, so that FinCEN and law enforcement can find the reports.

Activity Categories : For pig butchering, the alert asks filers to select "Fraud – Other" (field 34(z)) with the description "Pig Butchering", alongside any other categories that apply.

Granular Technical Documentation : The SAR narrative must meticulously record all known forensic indicators:

Beneficiary cryptocurrency public wallet addresses and blockchain networks.

Specific blockchain transaction hashes ( TXIDs ) and exact timestamps.

Fraudulent trading platform URLs, domain registrar details, and hosting IP addresses.

Communication channels (e.g., WhatsApp phone numbers, Telegram handles, dating profile links).

Asset Recovery: FBI IC3, RAT, and Stablecoin Contract Freezing

FinCEN Rapid Response Program (RRP) : For recent international wires, victims and institutions can report quickly to law enforcement (for example through the FBI's IC3), which can ask FinCEN to use the Rapid Response Program (RRP) to work with foreign FIUs to recover funds. Speed is critical, because the chances of recovery fall quickly after the first few days.

FBI IC3 Recovery Asset Team (RAT) : Filing an immediate victim complaint via the Internet Crime Complaint Center (IC3) , allowing federal investigators to coordinate with receiving correspondent banks to halt outbound clearing.

Stablecoin Smart Contract Freezes : Because major fiat-backed stablecoin issuers (such as Tether for USDT and Circle for USDC) maintain programmatic "blacklist" functions in their smart contract code, law enforcement coordination or formal intelligence reports can prompt the issuer to freeze the specific unhosted smart contract address holding stolen funds, locking the tokens on-chain before liquidation.

Comparison of Generative AI and Scam Attack Modalities

Attack Modality Underlying Enabling Technology Primary Target / Vulnerability Key Technical Signatures Primary Forensic & Compliance Defense

Synthetic Identity Creation LLMs (GPT/Claude); image diffusion models (Midjourney) eKYC document verification; credit underwriting Flawless document fonts; unassigned/deceased SSNs; no credit depth Non-documentary CIP verification; e-CBSV SSA matching; tradeline depth analysis

Deepfake Video Injection GANs (StyleGAN); virtual cameras (OBS); face-swapping Remote biometric selfie liveness verification (eKYC) Virtual camera drivers; frame edge jitter; absence of natural eye blink/rPPG Hardware attestation (Apple App Attest / Play Integrity); dynamic light challenge-response

Neural Voice Cloning Diffusion audio synthesis; zero-shot neural voice models Customer support telephone desks; voice biometrics Acoustic spectrum anomalies; unnatural breath pauses; metadata mismatch Out-of-band app push challenges; behavioral fraud scoring; FIDO2 MFA

Pig Butchering (Sha Zhu Pan) Social engineering; fake Web3 DApps; peeling chains Retail investors; elderly victims; crypto on-ramps Rapid HELOC/401(k) liquidation; immediate crypto transfer; peeling chains SAR with key term FIN-2023-PIGBUTCHERING; TXID tracing; IC3 reporting and FinCEN RRP for recent international wires

Realistic Exam Scenario

A transaction monitoring specialist at a regulated cryptocurrency exchange reviews an alert generated by a 67-year-old retail customer who opened an account 14 days prior. Over the past 72 hours, the customer initiated three separate inbound wire transfers totaling $350,000 from their domestic bank checking account, originating from a recently approved Home Equity Line of Credit (HELOC).

Upon each wire settlement, the customer immediately converted the entire fiat balance into Tether (USDT) on the TRON network and withdrew the tokens to an external, self-hosted unhosted wallet. When the exchange's customer support team placed a temporary security hold on the fourth outbound withdrawal of $100,000, a telephone call was received by the support desk from a caller claiming to be the customer. The caller's voice sounded authentic, but backend audio analytics flagged subtle unnatural spectral frequencies characteristic of neural voice synthesis. Furthermore, the customer's mobile app session revealed that the external wallet destination was interacting with a smart contract associated with an unverified Web3 decentralized application registered on an offshore domain only 12 days earlier.

Step-by-Step Compliance Analysis:

Typological Identification : The compliance officer recognizes the classic hallmarks of a Sha Zhu Pan (pig butchering) romance investment scheme: an elderly customer suddenly liquidating home equity, converting large fiat sums into stablecoins, and transferring them to a newly formed fraudulent Web3 trading application.

Emergency Asset Freezing : The compliance officer immediately escalates the case to senior management, asserting the institution's contractual terms of service to freeze the pending $100,000 withdrawal and lock the customer's account to prevent further losses.

Customer Intervention and Protection : Recognizing that pig butchering victims are deeply psychologically groomed and often deny that they are being defrauded, the compliance team initiates a secure out-of-band video verification call, tactfully explaining the indicators of investment scams, presenting blockchain analytics proving the funds were swept into a syndicate peeling chain, and guiding the customer to contact local law enforcement and adult protective services.

FinCEN SAR Filing (FIN-2023-PIGBUTCHERING) : Under 31 CFR § 1022.320, the exchange must submit a Suspicious Activity Report (FinCEN Form 111) within 30 days of detection. In accordance with federal guidance, the investigator:

Enters the key term FIN-2023-PIGBUTCHERING in field 2 and the narrative.

Selects "Fraud – Other" with the description "Pig Butchering", plus other categories that apply, such as elder financial exploitation.

Details all technical on-chain indicators: victim deposit addresses, the destination unhosted wallet public keys, the 14 individual blockchain transaction hashes ( TXIDs ), the fraudulent Web3 DApp URL, and the IP logs associated with the voice-cloned support call.

Law Enforcement Escalation and Asset Recovery : The compliance officer files a priority complaint with the FBI Internet Crime Complaint Center (IC3) , references the transaction hashes, and submits an emergency inquiry to the stablecoin issuer (Tether) to evaluate blacklisting the receiving unhosted wallet on-chain.

Common Exam Traps

Trap 1: Assuming Voluntary Customer Authorization Eliminates SAR Obligations : A frequent misconception is that if a victim willingly authorized outbound wire or cryptocurrency transfers, the institution has no legal basis to file a SAR. Under BSA SAR rules, and as FinCEN's pig-butchering alert explains, transactions that meet the SAR criteria must be reported even when the deceived customer authorized them.

Trap 2: Relying on a Selfie Upload Alone : Believing that a high-resolution selfie upload prevents synthetic fraud. Still photographs are easy to synthesize; robust eKYC needs independently tested liveness (passive or active) plus injection-attack defenses such as hardware attestation.

Trap 3: Omitting FinCEN Key Terms in SARs : Failing to include requested key terms such as FIN-2023-PIGBUTCHERING in pig-butchering filings. FinCEN relies on automated text indexing to aggregate and route scam data to specialized federal task forces.

Trap 4: Conflating Presentation Attacks with Injection Attacks : Assuming that 3D silicone mask detection software protects against virtual camera injection. Virtual cameras inject digital streams directly into application drivers, requiring cryptographic device attestation rather than optical lens analysis.

Loading diagram... Generative AI Threats & Pig Butchering Scam Laundering Architecture