Face biometrics development targeting tricky tasks

Face biometrics are a common theme running through the most-read articles of the week on Biometric Update, along with the impact that deepfakes and synthetic data are having on the field’s market and development. New algorithms from Innovatrics and Incode performed well in NIST testing and a new training dataset of synthetic identities seeks to bridge the realism gap. A report from Entrust and Onfido reveals the alarming frequency of deepfake-fueled fraud attempts and a new group including the ACCS, AVPA and Privately is looking into defending age assurance systems against deepfakes. Facial recognition is also the mechanism that will be relied on to keep Australian teens off of social media and make Europe’s border protection plans operationally feasible. Elsewhere, Jordan Burris delves into Socure’s advice for the incoming American government on digital ID.

Most viewed biometrics news this week

A group of technology and child welfare advocates in Australia waded into that country’s debate on banning social media to people under 16, arguing for the a more moderate regulatory approach. The dangers of social media for children are well documented, but the proposal would affect how everyone uses the internet, while not everyone understands what is being proposed or how it would work.

NIST published a new update to its 1:N FRTE track earlier this month, which shows particularly impressive gains with new algorithms from Innovatrics and Incode. Both algorithms cracked the top 10 in multiple categories, while familiar entries from NEC, Idemia, Paravision, CloudWalk, SenseTime and Megvii showed top results.

NIST also published guidelines for the U.S. federal government’s identity verification and access control system, which is based on ID cards. The new documents cover derived PIV credentials and federated PIV identity services. Both are final public drafts, and comments will be accepted until January 10.

Digital identity should be given critical infrastructure status in the U.S. and a national taskforce formed to advance it, Socure says among five recommendations for the new government. Jordan Burris tells Biometric Update in an interview that the critical infrastructure status is a natural extension of designations already on the books, and explains how leadership and collaboration can enable a leap forward for America’s defense against fraud.

Remote pre-enrollment may be the best way to resolve border-check bottlenecks with faster operations for the EU’s EES, but is running up against the requirement for fingerprint biometrics and concern about contactless fingerprint spoofing. There were key points from an industry roundtable held by eu-LISA, and featuring insights from numerous stakeholders including iProov and Mühlbauer.

CBP is using automated software testing to help maintain the security of biometrics and other sensitive personal data, and prevent another incident like the Perceptics breach. Even the automated tests themselves require robust data security, and they are likely to be used even more extensively as CBP migrates more systems to cloud environments.

World ID enrollments are back on in Brazil, and off in the Dominican Republic, where a consumer rights regulator has ordered their suspension. The company has wrestled with perception, from the physical appearance of its iris biometric-scanning orbs to the latest accusations of “abusive” contract clauses in the name of “privacy rocket science.”

The frequency of deepfake attacks has reached every five minutes, according to the 2025 Identity Fraud Report from Entrust and Onfido. The report also shows more than half of all document fraud in the digital channel involves digital forgeries, and the crypto industry is a relatively common target.

Deepfakes pose a potential threat beyond financial fraud as well, and stakeholders within the age assurance community have come together to launch the DefAI project to defend age verification against presentation and injection attacks. The project is run by the ACCS, AVPA and Privately.

Technology and law are both being turned against deepfakes, at predictably differing paces. Authologic’s digital ID aggregation business model gets a closeup from Forbes in the wake of its $8 million fundraise. The company’s founders believe malicious use of AI is a threat to civilization itself. Meanwhile in Minnesota, a court challenge to an anti-deepfake law has been led on a detour by an academic who filed a brief that appears to contain citations hallucinated by a large language model, according to The Verge.

Synthetic data holds the potential to help train better facial recognition systems, but the existing databases fall short of the realism needed to realize that potential, say researchers from Idiap. They built a database with nearly 400,000 face images of 20,000 unique synthetic identities, which they call Digi2Real, using foundational models, to address the realism gap.

Please let us know about any interviews, editorials, podcasts or other content we should share with the biometrics and digital identity community either in the comments below or through social media.

Article Topics

biometrics  |  digital ID  |  digital identity  |  face biometrics  |  week in review