Improving eKYC Accuracy: How Platform Corrects AI Algorithmic Bias

Ant International Wins NeurIPS Competition for Fairness in AI Face Detection

Ant International, a leading digital payments and fintech company, has emerged victorious in the prestigious NeurIPS Competition of Fairness in AI Face Detection. The company’s dedication to developing secure and inclusive financial services is underscored by this achievement, especially in light of the increasing prevalence of deepfake technologies.

The widespread adoption of facial recognition technology across various industries has brought to the forefront the issue of algorithmic bias in AI. Research conducted by the National Institute of Standards and Technology (NIST) has revealed that many commonly used facial recognition algorithms exhibit significantly higher error rates when analyzing the faces of women and individuals of color. This disparity is attributed to the lack of diversity in training data and the demographics of those involved in building and controlling mainstream AI platforms. Biased algorithms can result in the denial of financial services to large segments of the population, posing a security vulnerability.

The NeurIPS Competition, held in conjunction with the Conference on Neural Information Processing Systems, a renowned AI conference, tasked participants with creating AI models that excel in performance and fairness across various demographic factors such as gender, age, and skin tone. Ant International’s team surpassed over 2,100 submissions from 162 teams worldwide by accurately detecting 1.2 million AI-generated face images representing diverse demographic groups.

The winning AI model developed by Ant International combines a Mixture of Experts (MoE) architecture with a bias-detection mechanism. The system trains two competing neural networks—one focused on identifying deepfakes and the other challenging the first network to disregard demographic characteristics. This innovative approach ensures that the system learns to detect authentic signs of manipulation rather than relying on demographic patterns. The model’s training incorporated a globally representative dataset and real-world payment fraud scenarios to ensure scalability.

Dr. Tianyi Zhang, Ant International’s general manager of risk management and cybersecurity, emphasized, “A biased AI system is inherently insecure. Our model’s fairness is not only ethical but essential in preventing exploitation from deepfakes and ensuring reliable identity verification for all users.”

The technology behind Ant’s winning entry is now being integrated into its payment and financial services to combat the threat of deepfakes. The company boasts a detection rate exceeding 99.8% across all demographics in the 200 markets where it operates.

Ant International’s technology aids customers in meeting global Electronic Know Your Customer (eKYC) standards, particularly during customer onboarding, without algorithmic bias. This is especially crucial in emerging markets where financial inclusion faces challenges.

With a customer base of over 150 million merchants and 1.8 billion user accounts, Ant International is renowned for services like Alipay+, Antom, Bettr, and WorldFirst. The company prioritizes AI security in its operations, with AI SHIELD serving as a framework for risk management. AI SHIELD, based on AI Security Docker, helps mitigate vulnerabilities in AI services, such as unauthorized access and data leakage.

AI SHIELD supports a suite of risk-management solutions that offer comprehensive protection for financial transactions, including defenses against deepfake attacks and fraud. Ant International claims that Alipay+ EasySafePay 360 has reduced instances of account takeover in digital wallet payments by 90%.

(Image source: “abstract art of a beautiful portrait, solid shapes, geometric shapes, neotokyo colors, muted colors, pixar, artstation, greg rutkowski, samdoesart, ge” – public domain)

Explore AI and Big Data with Industry Leaders at AI & Big Data Expo

For those interested in learning more about AI and big data from top industry experts, the AI & Big Data Expo in Amsterdam, California, and London is a must-attend event. This comprehensive expo is part of TechEx and is co-located with other leading technology events. Visit here for event details.

AI News is powered by TechForge Media. Discover upcoming enterprise technology events and webinars here.