New generative AI models now pop up almost every week, and even unsophisticated fraudsters use them to generate convincing ID verification sessions at scale.
That shift makes defense-in-depth—layering several independent checks so nothing hinges on a single point of failure—more crucial than ever.
Today, we’re excited to announce powerful new security features that make it even harder for fraudsters to manipulate or bypass identity verification—while also streamlining the experience for trusted users. These enhancements are available to all Plaid Identity Verification (IDV) customers out of the box, with no additional integration required.
New deepfake biometric and synthetic ID detection models
As AI-generated media becomes increasingly indistinguishable from reality, staying ahead of synthetic fraud is critical. That’s why we’ve enhanced our identity verification with advanced machine learning to detect deepfake injections and synthetic camera feeds during both liveness and document checks. Stronger camera source integrity checks now block manipulated feeds before they reach your flow, while upgraded face detection models catch visual anomalies linked to AI-generated content. We’ve also added multiple layers of security to liveness and document verification, including stricter checks for document tampering and forgery, improved selfie-to-ID photo matching, and dynamic liveness detection that resists spoofing attempts like printed photos or screen recordings. These upgrades help ensure the person on camera is the person on the ID—delivering faster, more accurate identity checks and stronger protection for your users.
Detect duplicate faces to catch repeat fraud
Fraud rings often reuse the same individual or digital likeness across multiple accounts—sometimes with tampered IDs or synthetic documents. Plaid’s Facial Duplicate Detection feature helps stop this by cataloging faces and identifying potential duplicates across both selfies and document portraits. With an industry-leading false match rate of just 1 in 1 million, you get high precision without any added friction. And with flexible configuration options, you can choose to match across your entire user base or within specific use cases, making it easier than ever to detect synthetic identities, stop mule accounts, and prevent repeat fraud attempts.
Age estimation for smarter risk signals
Introducing a new biometric tool that estimates a user’s age during the liveness session. This powerful signal helps flag major discrepancies between a user’s stated age, their ID document photo, and their selfie, offering an important indicator of potential identity misuse or impersonation. Our machine learning models are trained on diverse datasets to perform reliably across a wide range of demographics and environments, ensuring consistent, equitable results without compromising the user experience.
Enhanced new ML-powered user flow options
We’ve introduced new capabilities that make it easier to tailor the identity verification experience based on real-time risk—delivering the right amount of friction at the right time. Powered by our Trust Index, these enhancements help you streamline verification for trusted users while tightening controls for higher-risk cases.
Risk-Based Escalations: Dynamically adjust the verification path based on user risk. High-trust users can move through with less friction, while higher-risk users are stepped up for additional checks automatically.
Selfie Re-Authentication: Re-verify users by comparing a fresh selfie against one from a previous session. Ideal for sensitive use cases like account recovery, right-to-work enforcement, or high-risk logins—ensuring the person coming back is the same one who originally signed up.
Trust Index Risk Check: Gain more control over decisioning logic with expanded Editor options. You can now fail a user not only when specific risk categories exceed set thresholds, but also when the overall Trust Index score signals elevated risk, even if individual checks don’t trip their limits.
By combining rich data sources and high-fidelity signals with the scale of our network, Plaid gives companies a faster, more secure way to verify identity and stop fraud at the front door. And for existing Plaid IDV customers, these enhancements are available now—your solution is only getting better, and you can start benefiting from these advanced protections today.