Synthetic data-sets provide a ground-breaking tool to help firms test and scale innovative solutions to combat fraud. The Financial Conduct Authority (FCA) and the City of London are delivering a synthetic data-set, focused on Automated Push Payment (APP) fraud solutions, to be hosted on the FCA’s permanent sandbox. 

In the modern digital landscape, data is the currency that buys clarity and mitigates risk. It drives innovation, decision-making, and propels sectors like fintech to unprecedented heights. One such evolution on the horizon is the advent of synthetic data. As the City of London Corporation continues to safeguard London and the UK as the leading fintech jurisdiction, understanding the vast potentials of synthetic data becomes imperative. 

What is synthetic data? 

Synthetic data uses advanced algorithms, to replicate the structure and patterns of real-world data without containing any of the original information. Imagine a highly accurate clone of a dataset, resembling the original (through alignment with data patterns) but not tethered to any individual's personal details, thereby preserving privacy. It strikes a balance between utility and confidentiality, ideal for a world where both are paramount. 

Why is synthetic data needed? 

99% of organisations agreed that data is important to their success. Gartner predicts that by 2024, 60% of the data used for AI and analytics projects will be synthetic. Data, once gathered painstakingly from the real world, is now being synthetically generated, bridging gaps, and providing a cost-effective, secure, and rapid solution. Synthetic data's diverse applications promise a range of use-cases across sectors. However, as start-ups and investors navigate this promising landscape, understanding its potential and challenges is paramount. 

There are many use cases to synthetic data. First, sharing real data poses privacy concerns. Synthetic data can be designed not to include any personally identifiable information, making it safe for sharing without violating privacy laws or norms. Second, in many applications, especially in deep learning, large amounts of data are needed to train accurate models. If only limited real data is available, synthetic data can be used to augment this data and improve model performance. Third, some events, like financial system failures, do not occur frequently enough to gather a significant amount of real data. Synthetic data can simulate these rare events, enabling better training of predictive models. Fourth, related to scenario development, it can be a cost effective, very specific, and more ethical tool for simulation of risky situations. This in turn can be improve model robustness. Finally, it can allow for tracking of how an algorithm can perform – whether an early prototype or something that it more developed. 

Using synthetic data to combat fraud 

In a pioneering collaboration, the FCA and City of London recently announced their synthetic data initiative centred around Authorised Push Payment (APP) fraud. APP fraud, an increasingly concerning financial crime, involves duping individuals into authorising a payment to an account controlled by a fraudster. The initiative's brilliance lies in its approach: by using synthetic data that mirrors real transactional behaviours (without jeopardising any individual's information), stakeholders can develop, test, and refine solutions in a realistic environment. APP Fraud losses are estimated at £485 million in 2022 alone, and finding effective technological solutions has become challenging due to the fast-paced changing nature of fraud.  

However, the FCA and City of London Corporation have continued to explore and enhance the synthetic dataset and are excited to announce this initiative to build on the FCA’s APP Fraud sprint and plot the role for synthetic data.  

The dataset will be hosted on the FCA permanent sandbox and available for innovators who apply to continue to ideate and build solutions. The Kalifa Review underscored the significance of innovation in UK fintech, including highlighting the need for a permanent digital sandbox. Sandboxes are controlled environments where businesses can test innovations without the constraints or consequences of real-world regulations. Start-ups and established entities alike can experiment, refine, and perfect their offerings, all under the vigilant, yet supportive, eyes of regulators. The flourishing of sandboxes is a testament to the UK’s commitment to fostering fintech ingenuity while ensuring robust oversight.