Public Funding for Fincrime Dynamics Ltd

Trustworthy AI is a set of principles and practices to ensure that AI systems are transparent, unbiased, and reliable. In financial crime analytics, AI-based solutions are increasingly used to detect fraudulent activities. However, using AI in financial crime analytics raises concerns about models' transparency, bias, and robustness. We implemented trustworthy AI concepts into our solution called Synthetizor, a tool for creating and deploying financial crime vaccines. The use of trustworthy AI in financial crime analytics is important because it can help improve the solutions' effectiveness and increase the trust of financial institutions and regulators in the solutions. Trustworthy AI can help to ensure that the models used to detect and prevent financial crime are transparent, unbiased, and robust and that the decisions made by the models are interpretable and explainable. It can increase the accuracy of the models, reduce the risk of false positives and negatives, and ensure compliance with regulations. Building a consortium for financial crime vaccines is an exciting opportunity for potential partners from academia and industry to collaborate and tackle a pressing problem collaboratively and innovatively. By working together, partners can leverage their strengths and expertise to develop cutting-edge solutions that address the specific needs of financial institutions. The project aims to conduct a feasibility study for building a consortium of partners from academia and industry to address the problem of financial crime using financial crime vaccines. The study will evaluate such a consortium's potential benefits and challenges, including the impact, required resources, and risks. We are building a consortium for trustworthy AI in financial crime on a study case from a pioneer organisation that is currently trying the first financial crime vaccine for automated push payments fraud. By studying a pioneer organisation's experience with implementing AI solutions using financial crime vaccines, the consortium can learn about the challenges and opportunities specific to the field and the best practices to improve the performance and trustworthiness of the solutions. We also have expressed interest from three universities and several financial institutions in the UK. The project's outcome will be a feasibility study that assesses the potential benefits and challenges of building a consortium to address the problem of financial crime using trustworthy AI. The study will recruit potential consortium partners and work with them to identify the problem and possible solutions.

During COVID-19, the financial behaviour of people changed. Crime evolved to a new level. We are starting to witness rising financial crime rates where there is no presence of the cardholder and abuse of government support for the crisis. These problems require new ways for control systems to rapidly adapt to this reality, which will demand a corresponding benchmark that can appropriately measure the performance of transaction monitoring systems, which is one of the biggest challenges financial institutions face today. The performance measurement of machine learning algorithms is usually done with metrics like precision, recall and others that derive from these values, such as F-score. Precision allows us to identify, from the whole set of criminal behaviour detected in a given dataset, how much of it is actually crime, whereas recall gives the number of real crimes detected in proportion from the total amount of crime present in the dataset. The biggest issue of relying on these metrics as benchmarks in fincrime analytics is that the exact amount of hidden crime present in the real dataset is unknown, effectively invalidating the reliability of these conventional metrics for fraud analytics. This project introduces CP-Mark, a benchmark for evaluating controls in financial institutions. Financial institutions tune their control systems according to applicable regulations, which carries two clear objectives: detect and prevent as much criminal activity (increasing true positives), and reduce the number of innocent people wrongfully accused (reducing false positives). Financial institutions' efforts to achieve these goals are hindered, mainly because of their inability to adequately assess the actual amount of hidden crime present in their datasets, rendering conventional metrics with little benefit. Criminals leave fingerprints of their activities in financial institution's records. Unfortunately, the use of this data is very restricted under privacy regulations such as GDPR, significantly reducing the possibility of collaboration between different stakeholders to improve fraud control tools and prevent financial crime. A crucial part of the solution to these problems will require a combined response of enriched synthetic dataset generation and proper metrics that can adequately benchmark performance and effectiveness of machine learning algorithms that operate as part of transaction monitoring systems. PaySim is a payment simulation software that creates digital synthetic data enriched for advanced solutions based on machine learning techniques to understand the patterns in data that lead to financial misbehaviours. These patterns are extracted from real data sources preserving its privacy constraints, capturing the dynamics of fraud and combining them into tailor-made scenarios of diverse crime typologies. This project will allow us to evaluate the use of Conformal Prediction (CP-Mark) as a reliable benchmark tool to test the effectiveness of our software and other Machine Learning algorithms used as part of transaction monitoring systems. We will then perform benchmarks on several controls using these datasets with a state-of-the-art machine learning framework called conformal prediction to build predictive models capable of detecting known and currently undiscovered patterns of fraud.

The importance for our society to reduce criminal profit and deter future generations from finding financial crime as a rewarding lifestyle is one of the key aspects that we are envisioning to contribute with this project. Criminals leave a fingerprint of their activity in the financial institutions on their financial records. Unfortunately, financial data is very constrained by customer privacy regulations such as GDPR. This hampers the possibility of collaboration between different stakeholders in financial problems such as optimising fraud controls tools and reducing financial crime. Solutions based on Machine Learning (ML) are starting to arise, but the quality data required to properly train the models is not available. With the current COVID-19 pandemic, crime has evolved to a new normal and so has the behaviour of non-fraudulent people. Evidence of this is the rise in digital ecommerce and the types of fraud where there is no presence of the card holder. These problems require new ways to rapidly adjust to this new normal. We address these problems using advanced financial simulation. Our innovation is called FRAUDSIM, and it enables enhanced and rapid deployment of machine learning for financial crime analytics. Our solution creates digital synthetic twins of financial data especially enriched for improving machine learning fraud controls. FRAUDSIM captures the fraud dynamics and combines them into tailor-made scenarios that explore diverse threats that financial organisations are exposed. Our simulator outputs augmented non-confidential financial synthetic data, resulting in trustable datasets ready-to-use for solution providers of advanced financial crime analytics. With FRAUDSIM, we will be able to rapidly develop, benchmark and compare advanced solutions based on machine learning across the industry. These enriched synthetic datasets are the new oil for ML engines to tackle not only the fraud control optimisation problem, but also a diverse set of complex problems currently present in the interaction between government regulators, financial firms, academia and third-party providers.