In Hollywood movies, all that a government agent needs to identify or track a person of interest is to simply click on their mouse a few times and immediately get all the details there is to know. In the real world, this would have taken many systems to work together seamlessly – not an easy task.
Today, biometrics technologies, such as face recognition, seem all too easy with cameras everywhere. Plus, the promise of AI to crunch incredible amounts of data suggests surveillance and homeland security systems can be all knowing.
The reality is a lot more difficult, says NEC, which supplied Singapore’s first biometric passports in 2005 and helps provide safer city technologies across the world today.
Key to a unified view of surveillance today is making all the systems work together, especially legacy ones that many governments have, the Japanese firm notes.
In this month’s Q&A, NEC Laboratories Asia-Pacific vice-president Christopher Lam and NEC’s head of delivery for homeland security Gene Tay tell Techgoondu that biometrics and AI are already working together in law enforcement and anti-terrorism efforts.
Equally important are questions of privacy, consent, retention policies, auditability and cybersecurity, they add, because the larger and more connected the system becomes, the more important it is to protect against misuse, unauthorised access or model drift.
NOTE: Responses have been edited for brevity and clarity.
TG: Biometrics and AI have the potential to be fused together for a number of uses, for example, in law enforcement or anti-terrorism efforts. Could you share some examples of how this has been deployed or can potentially be deployed?
Lam: Biometrics and AI can complement each other very effectively when applied to high-assurance identity and situational awareness use cases.
In practice, one of the clearest examples is border management, where biometric matching can help verify that a traveller is the rightful holder of a document, while AI helps analyse multiple data points in real time to improve throughput, flag anomalies and support frontline officers in making faster decisions.
Similar approaches can also support public safety operations, for example, by helping authorised agencies identify persons of interest from large volumes of video feeds, provided the deployment is governed by clear legal safeguards, strong oversight and defined operational purpose.
We also see potential in areas such as fraud prevention, secure access to critical infrastructure and digital identity services, where the fusion of biometrics, analytics and workflow automation can improve both security and user experience.
The key point is that these systems should not be viewed as surveillance for its own sake, but as targeted decision-support capabilities designed for specific, lawful and proportionate outcomes.
At Expo 2025 in Osaka, NEC deployed its facial recognition system across the venue to support both admission gate management and cashless commercial transactions.
When visitors scan their ticket QR codes, the system performs real-time 1:1 facial verification to prevent identity theft and ticket sharing.
For retail purchases within the venue, the platform securely matches facial captures against encrypted financial account information, enabling visitors to complete transactions without the need for physical payment cards or mobile devices.
Separately, working with Star Alliance and aviation technology provider SITA, NEC has deployed a biometric identity verification platform at Frankfurt and Munich airports to enable contactless and paperless passenger journeys.
Using NEC’s high-speed “1:N” facial recognition matching engine, it allows travellers to move through multiple airport touchpoints, including check-in, baggage drop, security screening and boarding, without presenting physical passports or boarding passes.
The underlying AI technology optimises facial matching speed and feature indexing to process large volumes of international passengers simultaneously while maintaining operational stability under varying environmental conditions, including differing lighting situations at airport gates.
The result is a smoother passenger experience and improved operational efficiency across airport processes.
TG: For governments, what are some of the challenges of achieving this fusion of systems, given the many data sources, say, from thousands of cameras and immigration records?
Tay: For governments, the biggest challenge is not simply deploying face recognition, but integrating it with many other systems at scale.
Data may come from thousands of cameras, immigration databases, watchlists, and case-management platforms, so the key issues are interoperability, latency, data quality, and privacy governance.
Public-sector solutions must combine authentication, accessibility, compliance to ISO standards, interoperability, and data protection rather than operate as isolated tools.
Lam: Governments often need to integrate legacy systems, live video streams, immigration or identity databases, and operational workflows that were not originally designed to work together. That creates issues around data quality, interoperability, latency and governance.
Equally important are questions of privacy, consent, retention policies, auditability and cybersecurity, because the larger and more connected the system becomes, the more important it is to protect against misuse, unauthorised access or model drift.
Another practical challenge is scaling performance without overwhelming human operators. Thousands of cameras and millions of records can generate enormous signal and noise, so agencies need systems that prioritise actionable alerts, reduce false positives and fit naturally into operational decision-making.
In our view, success depends as much on systems integration, governance and human factors as on the biometric algorithm itself.
TG: There have been efforts to move biometrics from a centralised model where biometric data is stored centrally to one where a user’s device holds the biometric information. Would this be a more secure way forward to authenticate users digitally?
Tay: It is a balance act between user requirement, security and privacy. Matching at the edge device can be more secure in some way, but it is not necessarily safer, overall.
The main gain is that keeping biometrics data on user’s device reduces the risk of a large scale central breach target and improves privacy control, while the tradeoff is that device security, revocation, and interoperability becomes harder. If the device is compromised, the biometrics credentials can still be attacked locally, so endpoint security becomes critical.
Lam: Centralised and decentralised models each have a role, depending on the use case, assurance requirements and regulatory environment.
For consumer authentication, device-held biometrics can be highly effective when combined with strong cryptography, device integrity checks and fallback mechanisms.
For national-scale identity or border use cases, there may still be a need for centrally managed reference systems, but these should be designed with strict minimisation, template protection, access control and life-cycle governance.
So I would say the more secure way forward is not simply decentralisation by itself, but privacy-by-design architecture that places biometric data only where it is operationally necessary and protects it throughout the full system life cycle.
TG: Biometrics have come a long way, with accuracy and speed now greatly enhanced. What differentiates NEC’s biometric technologies, such as face recognition, from your rivals that have been enhancing features such as liveness in recent years?
Tay: NEC’s strength is broader. It combines world leading benchmark performance, repeatedly ranked first at NIST (National Institute of Standards and Technology) testing, in accuracy, speed and tolerance for difficult images.
NEC is highly robust to poor-quality inputs, including low resolution, angle changes, age changes, and partial obscured faces, which matter most in public-safety deployments than in ideal lab conditions.
More than 70 countries and regions have adopted our systems. NEC’s systems are built for large-scale operation, such as multiple camera feeds and millions of multi-modal matches in a minute, so the differentiation is not only algorithm, but the robustness and scalability of end-to-end deployment model. Proven track record and trust matter together.
Lam: Beyond accuracy, our strength is in tolerance to poor image quality, ageing, angle variation and other real-world factors that often degrade performance.
We also take a multimodal view of biometrics and identity, so face can be combined with other modalities where the use case demands stronger assurance.
On liveness specifically, it is an important capability, but it should be part of a broader anti-spoofing and trust framework rather than treated as a standalone differentiator.
In our view, customers increasingly need an end-to-end solution that combines recognition, liveness or proof-of-presence, systems integration, privacy safeguards, governance and operational fit.
That is where NEC’s long history in biometrics, our applied AI expertise and our experience supporting mission-critical deployments make a meaningful difference.
