HyperVerge Passes Deepfake and Face-Spoofing Tests, Claims Indian First
Independent evaluations by Australia’s BixeLab examined HyperVerge’s identity-verification technology against digital injection attacks and physical face spoofs.
HyperVerge has announced that it passed independent evaluations covering digital injection attacks and sophisticated facial presentation attacks. In an announcement published on 23 October 2026, the company described itself as the first Indian identity-verification provider to achieve the combination.
The development concerns technology used during digital customer onboarding, where a business must determine whether the person completing a verification check is genuine.
BixeLab’s public records confirm separate evaluations of HyperKYC SDK with the HyperVerge FAS system, its liveness backend. The laboratory published confirmation letters associated with reports dated 8 October and 12 October 2026.
The records provide evidence of performance under specified test conditions. The company’s “first Indian” claim is a separate statement about its position in the market.
What HyperVerge’s Deepfake Detection Milestone Covers
The announcement brings together two forms of identity-fraud testing.
Presentation attack detection, or PAD, examines attempts to deceive a biometric capture device by presenting a substitute for the genuine person. In facial verification, examples can include a photograph, a replayed image or a physical mask.
Injection attack detection, or IAD, examines attempts to introduce substituted biometric data into the verification process. This can involve manipulated images or videos entering the digital capture pathway.
| Evaluation | Main focus | Framework referenced |
|---|---|---|
| Presentation attack detection | Spoofing at the biometric capture device | ISO/IEC 30107-3:2023 |
| Injection attack detection | Manipulated or substituted biometric data introduced into the system | CEN/TS 18099 |
The two checks address different routes into a verification process. An organisation needs to understand both the image it receives and whether the capture process has been interfered with.
This is particularly relevant when an AI-generated face or altered video can resemble the person a business expects to see.
What BixeLab’s Public Results Show
The laboratory evaluated specified Android and iOS versions for presentation attacks, and specified Android, iOS and web versions for injection attacks.
Its published summaries report:
| Test measure | Reported result |
|---|---|
| Physical presentation attacks | No attack-classification errors across 150 tests |
| Genuine presentations in the PAD evaluation | Two classification errors across 50 transactions, equivalent to 4% |
| PAD result at participant level | Every participant completed at least one valid transaction |
| Digital injection attempts | All 630 tested injection transactions were rejected |
| Genuine presentations in the IAD evaluation | A 1.33% classification-error rate across 300 runs |
The genuine-user figures need careful interpretation. A participant successfully completing a valid transaction does not mean every attempt by that participant was accepted.
Similarly, rejecting the tested injection attempts demonstrates resistance within the evaluated setup. It does not establish that every possible deepfake or future attack will be detected.
These distinctions help explain the results without reducing them to a single accuracy percentage.
Understanding the Standards and Testing Levels
ISO/IEC 30107-3:2023 establishes methods for evaluating and reporting presentation attack detection. Its scope concerns attacks at the biometric capture device; it does not provide an overall security assessment of an entire identity-verification system.
BixeLab’s PAD confirmation letter uses the Level 3 label and explains that the assessment followed its internal Level C sophistication framework. That describes the laboratory’s testing approach.
CEN/TS 18099 is a European technical specification addressing biometric data injection attacks. It provides guidance on evaluating detection systems, including their treatment of legitimate users.
HyperVerge’s announcement describes the injection evaluation as IAD Level 2. The laboratory’s letter calls it Level Substantial, conducted in close alignment with CEN/TS 18099.
The letter also states that this injection evaluation was not an NVLAP-accredited outcome or a certification or endorsement by NIST or a government agency.
The significance is the documented independent evaluation of named products and configurations.
How to Read the “First Indian Company” Claim
HyperVerge’s announcement presents the combination of evaluations as an Indian first.
BixeLab’s public directory independently supports the existence of the company’s two test records. However, the directory contains evaluations that providers have agreed to disclose; it is not a complete record of every assessment conducted worldwide.
The strongest directly verifiable finding is therefore that HyperVerge has published laboratory evidence for both types of evaluation. The national-first description remains attributed to the company.
Why This Matters for Banks and Fintech Companies
Digital onboarding allows a customer to apply for an account or financial service remotely. That convenience also makes reliable identity checks important.
HyperVerge positions its technology for use across banks, lending businesses, insurers and other digital platforms. Its tools are intended to help identify impersonation during onboarding, authentication and transactions.
For a financial institution, the business challenge has two sides. Accepting an impersonator can create fraud exposure, while rejecting a legitimate customer can cause delays, extra reviews and abandoned applications.
Independent testing gives procurement and security teams a more concrete basis for assessing a supplier. They can examine what was tested, which product versions were involved and how both fraudulent and genuine attempts were classified.
The practical value comes from matching that evidence to the institution’s own customer journeys and operating conditions.
How the Technology Fits Into Identity Verification
HyperVerge offers deepfake detection alongside face matching, liveness detection and document verification.
These functions answer related questions. Face matching compares a person’s image with a reference image. Liveness and attack-detection checks examine whether the submitted biometric material represents an acceptable, genuine capture. Document checks add information about the identity evidence supplied.
The company offers integration through APIs and software development kits, allowing these capabilities to become part of an organisation’s onboarding process.
From an operational perspective, combining the checks can provide a fuller picture of a verification attempt. The result still depends on the configuration, the quality of the submitted information and how the business handles uncertain cases.
What the Milestone Means for HyperVerge
The evaluations give HyperVerge documented results to present when competing for identity-verification and fraud-prevention business.
For potential customers, the most useful evidence extends beyond whether a test was passed. It includes the kinds of attacks examined, the treatment of genuine users and the scope of the assessed implementation.
The next commercial test is how the technology performs across actual deployments. Relevant measures include fraud outcomes, customer completion rates and the amount of manual intervention required.
HyperVerge’s announcement brings attention to a practical requirement in digital identity: security assessments must consider both physical impersonation and manipulation of the digital verification process.
For more insights navigate: PressPire | Stories That Shape the Future
Read previous article here : nvs-03-launch-isro-navic-positioning/
Follow PressPire on Facebook | Instagram | LinkedIn | YouTube
PressPire | Stories That Shape the Future
