Liveness detection and biometric fraud: where authentication breaks

At a glance

What this is: NHI Mgmt Group's analysis, drawing on Ping Identity's blog post: Liveness detection strengthens biometric authentication by verifying that the person presenting a face, voice, or fingerprint is physically present, not using a spoofed or synthetic sample.

Why it matters: It matters because IAM teams using biometrics for onboarding, recovery, or step-up authentication need controls that detect deepfakes and other presentation attacks before account takeover or synthetic identity fraud succeeds.

Context

Biometric authentication can confirm that a user matches an enrolled face or voice, but it cannot by itself prove that the sample came from a live person. That gap becomes material when deepfakes, replay attacks, masks, and injection attacks can feed synthetic media into remote identity workflows.

For IAM programmes, the issue is not whether biometrics work in principle. The issue is whether the control stack can distinguish genuine presence from convincing impersonation during onboarding, account recovery, passwordless login, and high-value transactions.

Most organisations now understand the fraud risk, yet many still lack formal protocols for AI-generated audio and video attacks. That leaves liveness detection as a governance and verification layer, not a standalone answer.

Key questions

Q: How should security teams use liveness detection in biometric login flows?

A: Use liveness detection wherever a biometric result would unlock meaningful access, especially in onboarding, recovery, and step-up authentication. The control should confirm live presence before the system trusts the match. Teams get the best results when they combine it with device trust, risk scoring, and recovery controls instead of treating it as a standalone gate.

Q: Why do biometrics need more than a face or fingerprint match?

A: Because a biometric match proves similarity, not trustworthiness of the surrounding process. Strong programmes add liveness detection, anti-spoofing checks, and contextual signals so replay attacks or synthetic inputs do not pass as legitimate users. Biometrics should raise assurance, but they should never be the only basis for identity acceptance.

Q: What are the signs that liveness controls are too weak for current fraud tactics?

A: Common signs include repeated recovery abuse, suspiciously successful remote enrolments, high completion rates with poor assurance, and weak resistance to spoofed media during testing. If the control passes only in lab conditions but not against realistic presentation or injection attacks, it is not performing at the level the workflow assumes.

Q: Should organisations use active or passive liveness detection?

A: Choose based on risk and user friction. Active liveness is stronger in visible challenge scenarios but adds more friction, while passive liveness is smoother and better for high-volume or lower-friction flows. Most mature programmes use a hybrid model, reserving active challenges for higher-risk events and using passive checks elsewhere.

Technical breakdown

How liveness detection verifies live presence

Liveness detection checks whether the biometric sample comes from a live person at the moment of capture. It does this by analysing motion cues, depth, texture, and challenge-response behaviour, then comparing those signals against patterns expected from a real face, voice, or fingerprint. The control is designed to separate a live capture from a photo, screen replay, mask, or AI-generated synthetic media. In practice, it sits inside the identity proofing or authentication flow and acts as a fraud filter before the system trusts the biometric match.

Practical implication: treat liveness as a verification step before biometric acceptance, not as an optional cosmetic check.

Active, passive, and hybrid liveness methods

Active liveness asks the user to perform a visible action such as blinking, smiling, or turning the head. Passive liveness analyses the image or video stream in the background without extra user prompts. Hybrid approaches combine both, often using passive checks for low-risk activity and an active challenge for higher-risk events. The architectural trade-off is between friction and resistance to spoofing. Active checks are easier for users to understand but can create drop-off; passive checks are smoother but depend more heavily on capture quality and detection accuracy.

Practical implication: align the liveness method to risk tier, transaction value, and acceptable user friction.

Why biometrics alone fail against presentation attacks and injection attacks

Presentation attacks happen when an attacker shows a fake face, voice, or document to the camera. Injection attacks go a step further by feeding manipulated content directly into the capture pipeline. Both can defeat systems that only compare a sample against an enrolled template. Standards such as ISO/IEC 30107-3 exist because the failure mode is not identity matching, but trust in the input channel. That is why liveness becomes important in regulated onboarding, password recovery, and high-value actions where account takeover has direct business impact.

Practical implication: validate the input channel, not just the biometric match, wherever remote access decisions carry fraud exposure.

Threat narrative

Attacker objective: The attacker wants to impersonate a legitimate user convincingly enough to obtain access, recover an account, or authorise a high-value action.

1. Entry occurs when the attacker presents a spoofed biometric sample, such as a deepfake video, replayed recording, mask, or injected media stream, into the authentication flow.
2. Credential access occurs when the system accepts the biometric match without proving live presence, allowing the attacker to satisfy the authentication gate.
3. Escalation follows when the same bypass is reused in onboarding, recovery, or high-value transaction flows, extending the attacker’s reach into accounts or approvals.
4. Impact is account takeover, synthetic identity abuse, or fraudulent transaction completion in a workflow that should have rejected the sample earlier.

Breaches seen in the wild

• DeepSeek database exposure 2025: An unauthenticated DeepSeek ClickHouse database exposed over a million log lines with plaintext chat history and API keys in 2025.
• Poland ArcGIS password leak 2023: An ArcGIS login emailed in 2020 was published from stolen mail in 2023 and still worked, exposing Polish military and infrastructure maps.

Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Biometric matching is not identity assurance: A face, voice, or fingerprint match proves resemblance, not live presence. That distinction matters because remote attackers now have convincing synthetic inputs that can satisfy a matching engine while bypassing the trust condition the control actually depends on. IAM programmes that equate biometric success with user legitimacy are treating the symptom as the control.

Proof-of-presence is the missing governance layer: Liveness detection changes the question from 'does this sample match?' to 'is this sample being presented by a live subject right now?' That is a governance shift, not just a technical feature. The control closes a channel-level trust gap that traditional biometric enrolment and matching do not address.

Biometric fraud is increasingly a workflow problem: The highest-value failures occur where onboarding, recovery, and transaction approval are tied to the same identity proofing stack. That makes liveness relevant across the lifecycle, not only at login. Practitioners should view it as part of the evidence chain for identity decisions, especially where fraud loss and regulatory exposure intersect.

Deepfake resilience is now an identity policy issue: The article’s central concept is the live-presence gap, where a valid biometric is mistaken for a trustworthy one. That gap shows why policy, risk tiering, and device binding matter alongside detection engines. In practice, the stronger control is the one that limits what the biometric can authorise when the sample source is uncertain.

Standards matter because the attack surface is the capture path: ISO/IEC 30107-3 and related guidance exist because attack success depends on how the biometric stream is captured, relayed, and evaluated. That means assurance should focus on testing against realistic spoofing and injection paths, not only on lab accuracy claims. Identity teams need evidence that the control survives the environment it will actually face.

From our research library:

• 43% of security professionals are concerned about AI systems learning and reproducing sensitive information patterns from codebases, according to the State of Secrets in AppSec.
• Read next: Ultimate Guide to NHIs — Key Research and Survey Results

What this signals

Live-presence verification is becoming a baseline control: Biometric programmes that do not separate match quality from presentation authenticity will keep over-trusting synthetic inputs. The practical shift is to make liveness part of the decision boundary wherever remote identity proofing affects access or money movement.

Fraud resistance now depends on workflow design, not just models: The same biometric engine can be acceptable in a low-risk login and inadequate in a recovery or transaction path. That is why risk tiering, device binding, and step-up challenges need to be designed together rather than added as afterthoughts.

Challenge-response is only one layer of assurance: Attackers can increasingly automate what used to be manual spoofing, so the stronger programme is the one that combines liveness with identity verification and policy-driven escalation. That approach reduces the chance that a convincing sample becomes an actionable identity event.

For practitioners

• Define liveness requirements by use case Set different assurance levels for onboarding, recovery, and high-value transactions so the biometric control matches fraud impact.
• Combine passive checks with step-up challenges Use passive liveness for low-risk journeys and trigger active challenges when the transaction or account risk increases.
• Test for spoofing and injection paths Validate whether your current capture flow can resist deepfakes, replayed video, masks, and media injection before trusting biometric acceptance.
• Bind biometric use to broader identity signals Pair liveness with device context, identity verification, and risk policy so a single biometric event cannot authorise high-impact access on its own.

Key takeaways

• Biometric authentication can still be deceived when the system trusts resemblance more than live presence.
• The control gap shows up most clearly in onboarding, recovery, and high-value actions where fraud has immediate impact.
• Liveness detection is most effective when it is paired with risk policy, device context, and identity verification.

Key terms

• Liveness Detection: Liveness detection is the mechanism that checks whether a biometric sample comes from a real, present person rather than a spoof such as a photo, screen, or mask. In identity programmes, it is a core defence against presentation attacks and should be tested under realistic operating conditions.
• Presentation Attack: A presentation attack is an attempt to fool a biometric system with a fake face, replayed video, mask, or other synthetic artefact. In practice, the control fails when it measures resemblance alone, because the attacker’s objective is to pass as the real user without actually being that person.
• Injection attack: An attack that inserts synthetic or manipulated data directly into the verification flow rather than fooling the sensor itself. For identity programmes, this is a control-path problem, because the attacker may bypass the visible presentation layer and exploit the software decision point.
• Proof of Presence: A verification approach that aims to establish that a real person is actively participating at the moment of authentication. It goes beyond matching a stored trait and instead looks for live, context-specific evidence that resists replay, cloning, and remote fabrication.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM programme, it is worth exploring.