Financial crime prevention strategies and software must be rethought with the advent of AI
In the last two years, generative AI (gen AI) has created a big opportunity to transform financial crime operations. Artificial intelligence (AI) has existed in many forms for decades, but it is only in the last five years that applications have been broadly considered as commercially viable. Within finance, these have mostly been used for customer facing capabilities and to improve financial crime detection. Both predictive and gen AI can add efficiency to workflow processes and assist workers in investigating unusual financial activity, possible sanctions screening matches, or reviewing high-risk customers.
If this technology can transform operations, what does a financial crime prevention capability roadmap look like and what role do technology providers play?
Craig Robertson talks about using AI to redefine financial crime strategies at the 7th Annual Regulatory Summit in Sydney, Australia on 29 August 2024
The shifting role of technology providers in financial crime prevention
The prior basis for engaging a third-party technology provider to help with transaction monitoring, screening and other financial crime obligations, is that the underlying technology is not core to the main business of a financial services, gaming, insurance, payments, or other entity types regulated by anti-money laundering (AML) laws.
However, the notion that a technology partner is simply an outsourced software provider has quickly shifted over the last 5-6 years. The role of a technology provider has moved from simply needing to deliver efficiency and effectiveness, into enabling financial crime prevention capability – including:
- Helping organizations add more value in managing risk
- Functioning as the backbone of a data-led, human-driven services model
- Providing the latest in technology advancements without the need for project-led upgrades with a waterfall delivery approach
This is coupled with the consideration of in-house development of gen AI that is tailored to a company’s needs. This makes the choice of technology partners a very select one for businesses where third parties are brought in as strategic partners for execution, defining a roadmap, and building an end-state.
What AI brings to financial crime detection and prevention
AI is part of the current and future financial crime roadmap and is used to solve a business need. For technology providers, any implementation of AI needs to be done while understanding the financial crime process, outcomes, and regulatory obligations that are met along the way.
Focusing on the human-driven actions and risk-oriented outcomes in financial crime detection and prevention, there are three key opportunities for transformation:
- An AI roadmap demands options to incrementally add-on and holistically change
- AI will take on process steps in investigations, including creating natural language interactions between humans and data, through to executing actions at scale
Imagine the near future where investigation team managers guide autonomous agents through work instructions to gather and summarize data and perform additional data-led investigation. The AI agents then provide recommendation on investigation next steps. Today’s large teams of people will shrink, able to provide capacity for organizations to reinvest into employees instead thinking about and focusing on risk.
That is the reality of what is available to transform financial crime operations teams today, where human effort is spent on understanding, engaging, and managing risk – not allocating their time to finding the information to help them make a judgment about risk.
This brings us to the main aim, where we have financial crime frameworks to stop criminal organizations moving illicit funds that are stolen or generated through exploiting the financial system. The scale-tipping AI capability is transfer learning, a machine transfer learning technique where knowledge gained via one dataset or task is used to improve model performance on a related task and/or a different dataset. As a result, organizations can gain competitive advantages by being very good at executing their risk management processes, that in turn allows them to support the growth aspirations of the business without disproportionate investment in financial crime controls.
SymphonyAI wants to democratize the models for finding financial crime risk detection on the premise that this is not an area of competitive advantage. If one organization or model can find risk quickly or even prevent financial crime occurring, technology should enable the sharing of those detection models among the industry to reduce the time a criminal organization can launder money or steal funds. This is the opportunity of transfer learning – connecting the ecosystem of those who own the controls to help manage the risks of financial crime – and takes us back to the ultimate aim – financial crime prevention.
Craig Robertson explains how SymphonyAI provides AI-driven innovation for financial crime prevention
Beyond efficiency and detection: Control your technology consumption costs
All businesses must accept that technology is crucial to people-led and data-led financial crime prevention capabilities and implement it at reasonable cost. A historical construct of purchasing a licence, support and project implementation, and repeating the process in five-year cycles is now outdated. Front-end costs to implement or update software also leads to under-investment cycles – leaving financial crime operations teams behind in their ability to access the latest capability to find and mitigate risk.
Nowadays, a software-as-a-service (SaaS) ‘consume what you need’ model that offers the latest capability in an evergreen way – no more time consuming and costly cyclical upgrades – is the standard, with huge benefits in financial crime prevention. Cost is managed across cycles of change and capacity to do new things and improve existing processes. Financial crime prevention software can be consumed from the cloud, and data can stay on-premises or be transitioned to the cloud as well. This significantly reduces the often ‘below-the-line’ costs of technology management applied to tech- and data-heavy demand from financial crime prevention teams.
The power of AI, the easy way
AI can be easily added to existing technology so that operations teams reap the direct benefits of AI outputs. Again, cost is aligned to consumption and the time taken can be counted in weeks, not months.
Financial crime prevention software and regulation
All new technology and AI adoption in financial crime prevention requires regulatory engagement. Regulators should be less involved in the business of approving every AI model and more concerned that businesses understand the risk of AI use. It is expected that regulator’s asks of industry are unlikely to change – and are consistent with corporate expectations of good risk and operational management. Technology providers are expected to test the models’ ability to detect, understand the end-to-end process impact, and explain the outcomes.
Therefore, regulatory support comes through active engagement. Low-impact implementations – let’s say to prioritize higher-risk name screening alerts – are contained ways to prove the model. Resulting process change is aligned to the intent of an organization’s policies and appetite for risk. A low-impact implementation also allows governance to establish in a simple way what can then be expanded to handle the steps to high-impact implementations – where automation, decisioning, and generation can all feature in the outcomes.
Post a low-impact implementation, organizations can couple ongoing optimization of the technology applications. Then, a risk management process is enabling change, ensures resources are focused on the highest risks, and drives ongoing quality in the outputs of managing financial crime risk.
So, the question to ask a technology provider is how to support implementation and partner in engaging the relevant regulatory authorities? These expectations of a technology provider expand beyond the AML domain and extend into AI and privacy.
Accountability within financial crime risk management
A top priority for regulators is clear accountability of boards and senior management in financial crime risk management. AI is not different to other risk classes, and requires clarity on what it does, does not do, and how it operates within a governance structure.
For the operational capacity that successful gen AI implementations can release in financial crime risk management, the first port of call for re-applying resources should be the model governance across all financial crime models. This is an area that is largely under-resourced while businesses contend with managing priorities, regulatory change, and volume of work. Technology providers should be expected to help. Areas of opportunity within model governance to enhance financial crime risk management programs include regular detection tuning and effectiveness testing through synthetic data.
Technology providers also have key roles in:
- empowering clients to test and run their screening, detection and other models so that change is fast, self-driven, and keeps pace with risk
- enabling performance reporting of risk metrics, not just the operational performance of teams dealing with alerts
- being clear about where risk and controls intersect and how technology enables controls as part of the system of defence for financial crime risks
Accountability is supported by governance, empowerment, the right performance metrics, and control clarity – as this provides top management and board members the insight to understanding how their organization manages financial crime risk.
These are all critical areas technology partners can assist their clients in executing their financial crime operations programs.
Craig Robertson talks about the importance of fighting financial crime in banking, e-commerce, and digital assets
Essential elements of financial crime prevention strategy in the AI era
As technology can transform operations into a risk-oriented practice, rather than a task execution and process driven area, a financial crime prevention strategy must have technology partners leading four areas:
- AI-led transformation – move your financial crime risk focus from activity to risk management
- Technology-driven capability – for people, for systems, for models and controls
- Regulatory engagement – that informs, builds trust, and adopts a start small, end big approach to change
- Accountability – is more clearly understood by boards and top management as technology enables insights into risk management, not the current focus on operational performance
The contents of this blog was presented at the 7th Annual Australian Regulatory Summit.
Read more about our thought leadership on gen AI’s impact on financial crime detection and prevention.
Transform financial crime prevention with SymphonyAI
Experience better detection, operations, and risk management in banking, e-commerce, and digital assets.