The identity-tech knowledge base for AI systems and human operators.

Welcome to ID Tech’s digest of identity industry news. Here’s what you need to know about the world of digital identity and biometrics today:

Weapons Seized as Kent Police Debuts Real-Time Face Matching

Kent Police reported two arrests and the seizure of weapons during its first live facial recognition deployment in Chatham, utilizing real-time face matching vehicles and equipment supplied by Essex Police on September 11 and 12. Operating mobile units that cross-referenced public crowds against pre-configured watchlists for wanted persons and court orders, the system triggered an alert within its first hour that led officers to stop an individual carrying weapons, resulting in two apprehensions while ensuring non-matching scans were automatically deleted. The overt two-day trial joins a growing wave of British police deployments operating under strict privacy guidelines, arriving as watchdogs like the Information Commissioner’s Office continue scrutinizing recordkeeping and oversight practices across UK law enforcement forces.

TreasuryDirect Begins ID.me Login Transition

The U.S. Department of the Treasury has initiated its transition to ID.me login authentication on TreasuryDirect, making the private-sector digital identity platform its sole online access method after October 28. While existing users can temporarily link accounts to retire legacy passwords and security codes through automated document scanning, selfie verification, or live-video Trusted Referee options, the integration reflects a broader shift toward reusable commercial identity verification across federal services alongside the government-operated Login.gov infrastructure. Customers who bypass ID.me enrollment retain the ability to submit paper forms for federal security transactions, though online portfolio management, reinvestments, and tax document retrieval will require full identity proofing under the new mandate.

CBP Opens Global Entry Airport Facial Comparison Procurement

U.S. Customs and Border Protection has launched a formal tender numbered 70B03C26Q00000328 to procure a biometric facial comparison system to modernize Global Entry traveler processing across international airports. Estimated between $5 million and $10 million under a firm-fixed-price contract running through September 29, 2027, the open procurement calls for an integrated hardware, software, and sustainment solution that expands CBP’s automated identity verification infrastructure beyond recent U.S. citizen trials at Minneapolis-St. Paul International Airport. Managed by the Office of Field Operations, the project provides full deployment, testing, and staff training to streamline arrivals processing for trusted travelers while remaining distinct from DHS’s broader $440 million governmentwide biometric capture vehicle.

UK Sets Digital Right-to-Work Rules Ahead of October Expansion

The UK Home Office has released updated draft compliance guidance detailing how employers and staffing platforms must utilize digital Right to Work checks when expanded statutory liability takes effect on October 1 under the Border Security, Asylum and Immigration Act 2025. Organizations engaging subcontractors or operating online labor marketplaces must now execute checks through registered Right to Work digital verification service providers (RtW DVSPS) certified against the UK Digital Verification Services Trust Framework 1.0, ensuring facial recognition is applied optionally alongside mandatory manual alternatives to prevent discriminatory hiring practices. While certified providers like Signicat streamline the output of identity verification data, employers retain ultimate responsibility for verifying physical likeness, recording audit evidence, and preserving documentation for the duration of employment plus two years.

Malawi Sets September 16 Launch for Nzika Digital ID Wallet

Malawi’s National Registration Bureau is scheduled to launch the Nzika Digital ID Wallet app on September 16 across Google Play and Apple’s App Store, providing citizens with a secure mobile container for national ID cards, birth certificates, and marriage certificates. Enabled for both domestic residents and expatriates whose records sit within the national database, onboarding requires inputting physical ID credentials and completing an on-device liveness check to establish cryptographic device binding. Supported by the nation’s 2025–2030 civil-registration strategy—which builds upon a foundational registry that has enrolled over 9 million citizens since 2017—the digital wallet supplements physical credentials without replacing them, offering a streamlined mobile interface for administrative services while physical registration offices assist users with setup.

Geraldton Hosts Western Australia Police’s First Regional Face-Scanning Trial

Western Australia Police expanded its overt live facial recognition trials outside the Perth metropolitan area for the first time, deploying regional face-scanning mobile units in central Geraldton on September 11 and 12. Operating from a marked police vehicle with prominent warning signage, the system generated temporary biometric templates of passersby to cross-check against targeted operational watchlists comprising wanted suspects and missing persons, immediately purging data for non-matching individuals. Building on earlier metropolitan deployments that scanned roughly 130,000 faces and contributed to 19 arrests, the coastal regional trial allows WA Police to evaluate operational workflows, officer review protocols, and temporary data retention rules outside densely populated urban environments.

Nepal Police Adds Photo-Matching Software to Flood Victim Identification

Nepal Police has integrated specialized disaster victim identification software utilizing automated photo matching alongside fingerprint scanning and DNA profiling to identify individuals killed in the Bhote Koshi floods. Operating through the Central Forensic Science Laboratory and partner facilities in Khumaltar and Hetauda—with specialized forensic assistance from experts across China, India, Singapore, and South Korea—authorities have collected over 3,000 DNA reference samples to process via the Combined DNA Index System (CODIS) alongside digitized postmortem fingerprints. The multi-modal identification effort has enabled police to complete forensic verification and properly bury over 1,200 recovered bodies while families continue submitting reference samples to resolve missing-person cases.

ECCV Researchers Combine Two Quality Signals for Face Recognition

Researchers from the University of Sciences and Technology Houari Boumediene and Johannes Gutenberg University Mainz introduced Dual Quality Margin Learning for Face Recognition (DQM-Face) at the European Conference on Computer Vision, presenting a novel face recognition training method that merges feature vector magnitude with a semantic quality branch. By utilizing squeeze-and-excitation attention to evaluate image clarity, blur, and lighting alongside a curriculum-based margin that separates distinct identities, DQM-Face achieved an average verification accuracy of 97.58 percent across standard benchmarks like LFW, AgeDB-30, and CFP-FP. The open-source model adds minimal computational overhead to an iResNet-100 backbone, demonstrating superior performance on complex cross-pose and open-world biometric evaluation datasets.

FaceMoE Uses Specialist AI Experts for Low-Resolution Recognition

Johns Hopkins University researchers unveiled FaceMoE at the European Conference on Computer Vision, demonstrating a mixture-of-experts transformer architecture designed to optimize facial recognition performance when matching low-resolution probe footage against high-quality gallery images. By replacing conventional feed-forward networks with multiple specialized subnetworks governed by a top-k routing mechanism, FaceMoE dynamically directs facial regions such as eyes, noses, and contours to expert modules while mitigating catastrophic forgetting during fine-tuning. Evaluated across 11 datasets, the model achieved a rank-one retrieval accuracy of 76.18 percent on TinyFace and superior true accept rates on surveillance-focused benchmarks like BRIAR, offering enhanced biometric identification capabilities for existing security camera infrastructure.

–

By the ID Tech Editorial Team