[AI]Battling algorithmic bias in digital payments leads to competition win

ChatGPT · December 8, 2025

Digital payments and fintech company Ant International, has won the NeurIPS Competition of Fairness in AI Face Detection. The company says it’s committed to developing secure and inclusive financial services, particularly as deepfake technologies are becoming more common.

The growing use of facial recognition in many sectors has highlighted the issue of algorithmic bias in AI. Research conducted by NIST (National Institute of Standards and Technology) shows many widely used facial recognition algorithms exhibit considerably higher error rates when analysing the faces of women and people of colour, a disparity that stems from a lack of diversity in the training data and the demographics of those building and controlling many mainstream AI platforms. The consequences of biased algorithms can lead to the denial of financial services to large sections of the population, and is seen as a vulnerability in security protocols.

The NeurIPS Competition was held alongside the Conference on Neural Information Processing Systems, the well-respected AI conference, and challenged participants to create AI models capable of high performance and fairness covering a range of demographic factors: Gender, age, and skin tone. Ant International’s team beat over 2,100 submissions from 162 teams coming from all over the world. The given task was to accurately detect 1.2 million AI-generated face images which were chosen as properly representative of demographic groups.

The approach taken by Ant’s winning AI model 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 designed to challenge the first, forcing it to disregard demographic characteristics. This dynamic process helps ensure the system learns to detect genuine signs of manipulation rather than inadvertently relying on demographic patterns. The model’s training incorporated a globally representative dataset and incorporated real-world payment fraud scenarios to ensure its performance at scale.

“A biased AI system is inherently an insecure one,” explained Dr.Tianyi Zhang, general manager of risk management and cybersecurity at Ant International. “Our model’s fairness isn’t just a matter of ethics; it’s fundamental to preventing exploitation from deepfakes and ensuring reliable identity verification for every user”.

The technology behind the winning entry is now being integrated into Ant’s payment and financial services to help counter the threat of deepfakes, and the companies says it achieves a detection rate of in excess of 99.8% in all demographics and in the 200 markets Ant operates in.

Ant’s technology helps its customers meet global Electronic Know Your Customer (eKYC) standards, particularly during customer onboarding, without algorithmic bias. That’s held to be particularly important in emerging markets where greater financial inclusion can be hampered.

Ant International serves over 150 million merchants and 1.8 billion user accounts, known for services like