The payments industry is experiencing rapid change with emerging regulations, new reporting requirements, and high consumer demand. While this presents new opportunities and strengthens market integrity, payment service providers must implement the changes sustainably and minimise risks. This is particularly important in the payments sector, which is characterised by a mix of smaller, dedicated payments institutions, and larger banks that are also payment service providers.
AI and machine learning are essential elements in the toolkit, allowing smaller organisations to compete in a challenging, and increasingly crowded market.
Lack of awareness of the available tools is a key barrier to adoption of emerging technologies and AI in payments. This hinders further discussion on potential use cases within the payments industry, and subsequent decisions over the specialist expertise needed. Some of the most popular tools are listed below.
This has become a business mainstay for customer support chatbots, but the use cases are growing by the day. Innovative financial services firms use it to support regulatory reporting, code translation and internal manuals, among others.
These programmes can assess spoken or written language, typically supporting customer service and risk management functions.
As a subset of AI, these algorithms learn from and analyse data to detect patterns that are indistinguishable during manual review. It can be a standalone tool to support modelling, or an integral component of generative AI and natural language processing.
This draws on statistical techniques and is integral to forecasting and predictive modelling for improved financial and risk management.
In short, AI and other emerging technologies can reduce the need for manual oversight, improve accuracy, and deliver business insights.
False declines can create poor customer experience, with potential reputational and financial damage to the merchant. To address this, payment service providers can apply AI and machine learning to support smart routing for greater payment authorisation rates. This can assess millions of previous transactions and multiple data points to identify the most favourable digital path for payment. AI can also analyse each transaction and automatically assign any applicable exemptions to support Strong Customer Authentication, with minimal delay.
Building on the above, effective know your customer (KYC) processes can help payment service providers reduce the potential for failed payments. By reviewing complex data sets, AI allows firms to automate customer identification, track behaviours and conduct reliable KYC risk assessment. Generative AI can also enhance due diligence processes, summarise customer data from a broad range of sources, and streamline onboarding. This supports regulatory compliance, improves accuracy and makes good use of available resources compared to manual due diligence processes.
AI and machine learning can support Confirmation of Payee processes to reduce the potential for authorised push payment (APP) fraud. In addition to real time verification, machine learning can enable real-time transaction monitoring by analysing transaction patterns and applying behavioural analytics to flag anomalous transaction requests. Payment service providers can then leverage AI to flag high-risk transactions and automatically request additional multi-factor verification for approvals. Feasibly, AI in payments could also support the investigation, and processing, of mandatory APP reimbursement claims. This includes assessing claim veracity and whether a customer may have been complicit in the fraud or otherwise been negligent.
Malicious cyber actors increasingly rely on AI to boost the volume, and sophistication, of attacks. Payment service providers can improve their cyber security posture with real-time threat and anomaly detection. This includes use of machine learning to monitor network traffic, and user and entity behaviour analytics (UEBA) can identify potentially compromised accounts. AI can also assess behaviour on end-user terminals to detect any unusual or suspicious activity.
In addition to detection and monitoring activity, natural language processing can also assess online sources for improved horizon scanning. This includes everything from cybersecurity reports, to industry databases, to dark web forums, among others. Taking this a step further, generative AI can assess past incidents to establish attack vectors, and assign threat risk scores for improved cyber risk management.
The cost of regulation can be high and payment service providers are subject to stringent capital requirements, operational resilience, Consumer Duty, and safeguarding rules; not to mention a raft of regulatory returns covering everything from operational and security risk, to open banking interface performance.
Machine learning and generative AI can streamline reporting processes by drawing on structured and unstructured data from across the firm. This reduces the need for manual input and improves consistency across returns.
Crucially, machine learning and AI in payments can track compliance activity from the source regulation, through to the associated actions. This safeguards compliance activity as the business evolves and real-time dashboards can flag exceptions for early intervention and breach reporting.
AI is now a business staple, but for many payment service providers it’s tricky to know where to start. First, it’s essential to identify key areas of the business that rely heavily on manual processing. For some, this may cover broader operational elements such as regulatory reporting, or customer service. Others, particularly larger banks or more established payment service providers, may already make good operational use of emerging technologies as a shared service. These firms could benefit from a more targeted approach for optimising payment flows or reducing APP fraud.
Next, firms need to establish the underlying tools and skillsets needed to implement change. The team needs to include technical skills sets covering AI, generative AI and machine learning for optimum implementation. It’s also essential to include transformation, risk management and regulatory specialists to ensure safe adoption while protecting consumers. Good training is vital to ensure teams can safely apply the technology, with a healthy risk culture around its use.
It’s also important to remember that application of emerging technologies or AI in payments isn't a one-and-done activity. These approaches need continual refinement, with ongoing horizon scanning to meet emerging threats and remain competitive in a tough market.
For insight and guidance on the use of AI in payment services, get in touch with our team: Paul Olukoya, Paul Staples or Alison Kopra.
