Financial services across Southeast Asia are becoming increasingly digital. Customers can now open bank accounts, apply for financial products, and access payment services without visiting a branch. But this shift also brings new challenges in verifying customer identities remotely.
One major challenge is identity fraud. Fraudsters can use stolen identity documents, photos, recorded videos, or manipulated faces and deepfakes to impersonate legitimate customers and bypass digital onboarding systems.
As a result, matching a face with an identity document is no longer enough. Financial institutions must also verify that the person is physically present and is not using manipulated images or videos.
This is where liveness detection plays an important role in biometric identity verification.
Liveness detection is a biometric security technology designed to determine whether a person undergoing facial verification is physically present, rather than an image, recording, or other form of facial spoof.
During digital identity verification, users typically capture an image or video of their face through a smartphone or computer camera. Without additional security measures, fraudsters may attempt to trick the system by presenting a photograph, replaying a recorded video, or using manipulated facial images such as deepfake.
Liveness detection helps identify these presentation attacks by analyzing signals and characteristics that indicate whether the face being presented belongs to a live person. By checking for signs of a genuine human presence, liveness detection adds another layer of protection to facial verification.
Facial recognition or face match and liveness detection serve different but complementary purposes.
Face match compares a person's facial characteristics against an existing reference image, such as the photograph on an identity document. Liveness detection, meanwhile, checks whether the face being presented comes from a live person rather than a spoofing attempt.
Some common facial spoofing techniques include:
1. Photo-based spoofing
Fraudsters may present a printed photograph or an image displayed on another device to impersonate a legitimate customer. Liveness detection helps identify these attempts by assessing whether the face exhibits characteristics of a live person.
2. Video replay attacks
A recorded video of another person may be played in front of a camera to simulate a live verification session. Liveness detection can analyze relevant visual and behavioral signals to distinguish a genuine user from a replayed recording.
3. Deepfake
Deepfake can be used in attempts to impersonate individuals during remote identity verification. Liveness detection helps identify signs of facial manipulation and presentation attacks.
These capabilities are particularly relevant to fintech companies, where remote identity verification is often part of the process of opening accounts and accessing financial services.
Liveness detection is important for biometric verification, but it cannot prevent identity fraud on its own. Fraud can also happen beyond onboarding, at every stage of the customer journey. Each stage requires different layers of protection.
Onboard. This is where fraudsters try to create a fake identity. Liveness checks, document verification against government records, and AML screening stop deepfake selfies, forged IDs, and duplicate identities before the account exists.
Authenticate. After an account is open, the main threat is account takeover. A fraudster enrolls a new device, often after coaching or stealing credentials. Strong authentication asks whether the new device is trusted, checks whether the old device approves the change, and confirms with a live face match.
Transact. Even a verified customer can be targeted. Mule accounts, scam payments, and money laundering happen at the transaction level. A fraud and AML layer watches the payment itself, so suspicious activity can be stopped before the money leaves.
Each stage needs a different proof. Onboarding proves the person exists. Authentication proves the person is still the same one. Transaction monitoring proves the action is legitimate.
So, what financial institutions should do now?
At VIDA, we think about identity as a chain. We verify the person once at onboarding, using liveness and document checks against the government record. We then prove it is still them at each subsequent transaction using device-bound authentication (Phone Token) and biometric liveness detection (Face Token). For high-risk activity, we add risk-based authentication so that low-risk actions stay smooth and high-risk ones get stronger proof.
For account takeover, ID Fraud Shield, powered by Sardine, which identifies risky login and enrollment attempts before they succeed. Our goal is simple: a fraudster who gets past one check should still be stopped at the next.
Ultimately, preventing identity fraud requires more than verifying a face during onboarding. Financial institutions need to establish trust at every stage of the customer journey, from verifying who someone is, to ensuring the right person accesses the account and that every transaction is legitimate.