NEC on Biometrics and AI: Integration Is Harder Than the Movies Make It Look

NEC, the Japanese technology firm behind Singapore's first biometric passports in 2005, says the real challenge in deploying AI-powered biometrics isn't the algorithms — it's getting legacy systems to talk to each other at scale. In a wide-ranging Q&A with Techgoondu published Monday, NEC Laboratories Asia-Pacific vice-president Christopher Lam and homeland security delivery head Gene Tay walked through the practical hurdles of fusing facial recognition, surveillance data, and AI analytics into systems that governments and enterprises can actually depend on.

The core tension, according to Lam and Tay, is between what AI promises and what current infrastructure can deliver. Thousands of cameras, immigration databases, watchlists, and case-management platforms were rarely designed to interoperate, creating problems around data quality, latency, and privacy governance that no amount of model fine-tuning can fix. NEC, which has deployed biometric systems across more than 70 countries and holds top rankings in NIST accuracy testing, argues that success depends as much on systems integration and human factors as on the recognition algorithm itself. At Expo 2025 in Osaka, the company demonstrated its facial recognition platform for both admission gate management and cashless transactions, while its partnership with Star Alliance and SITA has enabled contactless passenger journeys at Frankfurt and Munich airports.

The interview also explored the device-held versus centralised biometrics debate, with both executives advocating a balanced approach: consumer authentication may work well with on-device matching, but national-scale identity and border applications will continue to need centrally managed reference systems backed by strict privacy-by-design architecture. On liveness detection and anti-spoofing, Lam noted that those capabilities should be treated as components of a broader trust framework rather than standalone differentiators, especially as AI-generated deepfakes become harder to detect.

Why it matters for Singapore: NEC has been a fixture in Singapore's identity infrastructure for two decades, from biometric passports to hospital face recognition trials with SingHealth, and both Lam and Tay are Singapore-based executives shaping the company's Asia-Pacific strategy. As Singapore pushes toward a unified digital identity ecosystem with Singpass as the backbone, the integration challenges NEC describes — legacy system interoperability, privacy governance, operator overload — are exactly the problems GovTech and IMDA are wrestling with right now. The Q&A serves as a reminder that Singapore's AI ambitions in security and identity will be constrained not by how smart the models are, but by how well the underlying systems talk to each other.