January 21, 2025
The U.S. Department of the Treasury’s Financial Crimes Enforcement Network (FinCEN) has issued a landmark alert — a growing threat powered by generative AI targeting financial institutions (FIs). This marks a critical acknowledgment of the risks reshaping the fraud landscape.
At iProov, we’ve collaborated with top U.S. banks such as and leading global FIs like UBS, ING, and Bradesco to fight deepfake-enabled fraud and have shared our insights on this arms race at FinCEN events. Our solutions, rooted in advanced biometrics with liveness detection, address the complexities of this rapidly evolving threat.
The New Reality of Fraud: Deepfakes and Generative AI
Deepfakes – synthetic media created using generative AI – are now a major tool for fraudsters. FinCEN’s alert emphasizes their use in forging hyper-realistic ID documents, spoofing video and audio, and bypassing traditional verification methods.
These developments align with insights from iProov’s threat intelligence team, which reveal evolving criminal tactics like blending real and fake PII into synthetic identities and using deepfake-generated accounts to enable money laundering. We are iProov one of the only companies named a Synthetic identity fraud (SIF) mitigation vendor by the Federal Reserve.
Key findings from iProov’s 2024 Threat Intelligence Report:
- 704% growth in face swap attacks in the second half of 2023 compared to the first half.
- 255% increase in digital injection attacks targeting mobile platforms.
The rise of Crime/deepfake-as-a-service marketplaces has also made fraud tools easily accessible. Increasingly accessible tools enable criminals to execute attacks at scale, manipulate metadata, and obscure detection.
FinCEN’s Key Red Flags – And How to Address Them
FinCEN outlined key red flags indicative of deepfake fraud, providing foundational guidance for institutions. These red flags align closely with the patterns we’ve observed, and we believe biometrics play a crucial role in addressing them.
Below, we’ve selected examples to demonstrate how science-based biometrics complement and enhance FinCEN’s framework:
FinCEN’s Red Flag #1: Inconsistent identity documents submitted by customers.Deepfake-generated documents may contain subtle visual or informational inconsistencies. |
FinCEN’s Red Flag #3: Technological Glitches or Suspicious Verification Tactics.A customer uses a third-party webcam plugin during a live verification check or changes communication methods during a live verification check due to excessive or suspicious technological glitches. |
Insight: Here, biometric verification enhances detection by binding a user’s face to their legal identity documents with unmatched accuracy, detecting discrepancies even manual reviewers would miss. This ensures genuine presence and blocks deepfake attempts before they succeed. Using biometric documents, which contain the photo of the user – digitally signed by a government authority – is a critical enabler of authenticity during onboarding.
If a fraudster tried to onboard at an FI with iProov using a fake identity document, they would still need to scan their real face (which wouldn’t match the document) or use a deepfake. In either case, the technology rejects the attempt. iProov partners with document verification specialists, scanning the document itself for anomalies and ultimately delivering multi-pronged defense. As fraudsters are constantly refining their tactics, a multi-pronged defense is essential. iProov combines biometric verification, document anomaly detection, and dynamic liveness checks to address threats like deepfakes and SIF from multiple angles – ensuring fraud prevention systems are robust enough to protect users at every stage of the verification process, across the whole user lifecycle. |
Insight: Signals such as using third-party webcam plugins can indicate that a user is leveraging a digital injection attack with synthetic media. This method, where deepfaked media is inserted directly into identity verification systems, is five times more frequent than persistent presentation attacks like photos or masks.
FinCEN indicates that fraudsters will attempt to avoid live person-to-person verification checks by triggering technical issues. But this implies that humans can be relied on to spot deepfakes, so fraudsters avoid checks. Our experience indicates bad actors will purposefully fail our liveness checks so that they are redirected to a live human – as they know human eyes are easier to bypass with live deepfakes than sophisticated biometric verification checks that deliver more security and higher identity assurance. |
FinCen Red Flag #5 & 6: Identity imagery flagged by online database or deepfake detection software.
|
Red Flag #8: A customer’s geographic or device data is inconsistent with the customer’s identity documents.Fraudsters often manipulate metadata, like location or device signatures, to mask their activities. |
---|---|
Insight: iProov is a commercial deepfake detection software that delivers the highest level of assurance in user identity. Biometric onboarding and ongoing authentication ensure accounts remain tied to genuine users, reducing downstream fraud risks and resource expenditure.We catch deepfakes via our injection attack detection (IAD), presentation attack detection, and other algorithmic processes. All of this is supplemented by real-time threat detection – iSOC.By tying biometric data to trusted identity documents, institutions can prevent bad actors from New Account Fraud and Account Takeover Fraud. This means FIs have to spend less time and resources on manual/behavioral tracking if they can trust in genuine presence established by science-based liveness – that it’s the right person using the account in the first place. All of this takes the burden off the organization and automates deepfake detection. | Insight: iProov performs enhanced metadata testing; our backend algorithms analyze metadata for signs of spoofing. Certain discrepancies in the metadata would raise flags – these discrepancies often also indicate the use of emulators, virtual cameras, or rooting, so they are detected and blocked. |
FinCEN Response Analysis: Technological Solutions Required
FinCEN’s recent alert demonstrates growing regulatory awareness of the deepfake threat and provides critical baseline indicators for institutions at varying stages of fraud prevention maturity. However, these red flags need to be paired with advanced technological solutions to tackle the increasingly sophisticated tactics enabled by generative AI.
Advanced biometrics can reduce the reliance on human reviewers, who face an increasing burden as synthetic content becomes more convincing. By introducing tools like iProov’s Biometric Solutions Suite verification, financial institutions can move from a reactive stance to a preventive one – ensuring fraud is stopped before it happens.
Sophisticated biometric systems avoid placing an unrealistic burden on human reviewers to detect increasingly sophisticated synthetic content that’s functionally undetectable to the human eye and introduce an inherently preventative security posture that can scale with the growing threat of deepfakes. They act as the “lock” that stops fraud at its source.
FinCEN’s alert acknowledges synthetic media has evolved to the point where it’s “difficult to distinguish from unmodified or human-generated outputs.” Accordingly, financial institutions must implement robust technological solutions that can scale with the growing threat.
Building Resilience: The Way Forward
The stakes are high. The financial cost of deepfake-enabled fraud is enormous, but the erosion of trust in digital services could be even greater. FinCEN’s red flags provide a solid framework, but advanced tools like iProov’s BiometricSolutions Suite offer the scalability and assurance needed to truly safeguard financial systems.
Our Identity Crisis in the Digital Age report outlines clear steps financial institutions can take to combat deepfake threats:
Adopt Science-Based Biometrics
- Implement multi-frame liveness detection
- Use passive liveness technologies
- Ensure only genuine users gain access
- Gather real-time threat intelligence to detect and respond to emerging threats from the evolving threat landscape rapidly, rather than placing faith in static defenses.
Invest in Challenge-Response Mechanisms
- Create unpredictable, unrepeatable verification methods
- Eliminate replay and injection attack vulnerabilities
Leverage Continuous Threat Intelligence
- Monitor emerging threats in real-time
- Continuously update defense mechanisms
Deepfakes Financial Services Crisis: The Time to Act is Now
Deepfake fraud is no longer a distant threat – it’s here, and it’s escalating. Financial institutions need to act now by adopting advanced biometric solutions to complement the foundational guidance provided. There are proven solutions to ensure financial systems are equipped to combat sophisticated fraud tactics at scale.
For an in-depth understanding of the evolving fraud landscape, download our 2024 Threat Intelligence Report. | For more information on our solutions for FS: visit our Financial Services page | For a consultative look at your use case, pain points, and our solutions to your organization’s demands: book a demo