AI-driven deepfakes and injection attacks fuel global biometric fraud surge
Fraudsters are increasingly using generative AI to create deepfakes and digital forgeries that bypass biometric identity verification systems in injection attacks. According to Entrust's analysis of over one billion identity verifications globally, deepfakes now account for one-fifth of biometric fraud attempts, with injection attacks rising 40% annually and deepfaked selfies up 58% in 2025.
What happened
Fraudsters are using generative AI to commit new account fraud and bypass biometric-based identity checks, according to Entrust's 2026 Identity Fraud Report based on analysis of over one billion identity verifications across 30+ sectors and 195 countries between September 2024 and September 2025. Digital forgeries now comprise over a third (35%) of document fraud attempts, driven by the accessibility of modern editing tools and generative AI that enable creation of hyper-realistic replicas of identity documents.
Deepfakes now account for a fifth of biometric fraud attempts, with instances of deepfaked selfies increasing by 58 percent in 2025. Methods include synthetic identities (AI-generated faces not corresponding to real people), face swaps (replacing one person's face with another in recorded or live video), and animated selfies (adding AI-driven movement to static photos).
Deepfakes are most frequently used in injection attacks, where fake images or videos are fed directly into identity verification systems, bypassing the camera and live capture process. The frequency of such attacks rose by 40% annually, with virtual camera injections most common and often paired with device emulation techniques to deceive verification software.
Reported impact
Deepfake attacks disproportionately target high-risk financial services sectors, with 60% of deepfake-based fraud attempts aimed at cryptocurrency platforms, 22% at digital-first banks, and 13% at payments and merchants. In payments, 82 percent of fraud attempts target the authentication process.
National ID cards account for nearly half of all fraudulent document submissions globally, with driver's licenses making up another quarter. In 2025, fraudsters targeted Bangladesh national ID cards more than any other document from any country.
Index analysis
Fraud is becoming faster, more organized, and more commercially driven, with fraud facilities located globally running attacks around the clock. Fraud attempts peak between 2:00 am and 4:00 am UTC, when defenses in many regions are offline.
Fraudsters typically recycle the same fake information across multiple fraudulent documents in a 'rinse and repeat' approach. Older document systems that still have paper versions in circulation are most easily compromised.
Governance controls
The report emphasizes the importance of robust liveness detection to combat spoofs and deepfakes across the identity verification lifecycle. Adding liveness detection—whether passive or active—is identified as essential to staying ahead of adaptive threats.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.