Mathematics and AI Working Together to Tackle Global Digital Scams

This joint research initiative between the University of Warwick and Nanyang Technological University (NTU), Singapore addresses one of the most urgent digital challenges of our time. Online scams are growing rapidly in scale, sophistication, and global reach, affecting individuals, businesses, and entire financial systems. While many organisations hold valuable information that could help identify and stop these scams, strict privacy regulations and commercial sensitivities make it impossible to simply pool data. This project confronts that challenge by developing new ways for institutions to collaborate without compromising confidentiality, trust, or regulatory compliance.

The research is led at Warwick by Professor Carsten Maple of WMG and Professor Wang Huaxiong from the School of Physical and Mathematical Sciences, NTU. Professor Maple is internationally recognised for his work in cyber security, privacy, and threat modelling, with extensive experience translating research into real world impact across government and industry. His research focuses on understanding how cyber threats emerge across complex digital ecosystems and how secure, trusted systems can be designed to counter them. Professor Wang is a leading mathematician whose work underpins privacy preserving technologies, particularly in cryptography and trustworthy artificial intelligence. His research enables computation on sensitive data without revealing the data itself, bridging deep mathematical theory and practical innovation.

At the heart of the collaboration is a new approach to artificial intelligence that allows organisations to learn collectively while keeping data private. Each participating organisation, such as a bank, telecom provider, or digital platform, trains its own scam detection model using its internal data. Rather than sharing that data, the organisation encrypts key mathematical parameters from its model and submits only the encrypted information to a shared platform. Using a technique known as fully homomorphic encryption, these encrypted contributions can be combined to improve the overall model without anyone gaining access to the underlying information in readable form.

This process is coordinated through a global signal exchange that acts as a secure hub for scam intelligence. Once the combined model has been updated, it is returned to the participants, who can verify its integrity before incorporating it into their own systems. No organisation sees another’s data, and the platform itself never accesses unencrypted information. Making this process efficient, accurate, and secure at scale presents a major scientific and engineering challenge, and one that remains largely unexplored.

The potential impact of the research is significant. The project aims to deliver a practical scam detection model that balances strong predictive performance with the computational costs of advanced encryption. Beyond combating scams, the underlying privacy preserving framework has clear applications in areas such as financial crime prevention, healthcare analytics, and supply chain security. There is also strong potential for patentable innovations and real world deployment through industry partners, including major technology companies and global anti scam alliances.

Looking ahead, this collaboration lays the foundation for a long-term research programme in privacy preserving and trustworthy artificial intelligence. It is expected to lead to joint academic publications, shared experimental testbeds, and open source toolkits that support secure and ethical data collaboration. By combining Warwick’s expertise in cyber security and threat modelling with NTU’s strengths in mathematics, cryptography, and enterprise innovation, the project demonstrates how interdisciplinary research can deliver solutions to problems that matter globally.