AI has accelerated fraudsters’ efforts, increased the efficacy of their schemes, and allowed them to evade detection for longer. Yuval Marco, General Manager of Enterprise Fraud at NICE Actimize, hears from fraud prevention teams at the world’s largest financial institutions (FIs) about how AI is amplifying their challenges. Here’s what he had to share:
Scams and authorised push payment (APP) fraud: as FIs strengthened account takeover (ATO) controls and modernised payments systems, fraudsters pivoted to target the weakest links – the customers – scamming them out of hard-earned money with AI-powered schemes. NICE Actimize’s 2024 Fraud Insights report revealed that authorised fraud grew 22% by volume last year. Scams circumvent traditional authorised fraud controls, testing FIs’ fraud defences. Many FIs are concerned scams may impact their bottom line with changing reimbursement requirements and liability shifts.
Payments fraud: over the last year, the average amount per fraudulent transaction increased 43% to USD 3,222. In North America, check fraud grew, with the average loss per check deposit (genuine and fraudulent), increasing 14% YoY to USD 4.90. In Europe, fraud continues to build on faster payment rails, with the average fraud amount rising 13% YoY. Knowing new accounts face stricter controls, fraudsters age accounts and beneficiaries to boost payment fraud success rates.
Regulatory compliance: as scams grow and drive huge losses, regulators around the world are pressuring FIs to ‘make the customer whole’. With the UK PSR that came into effect in October 2024, the pending EU PSR, and the US Protecting Consumers Against Payments Scams Act, regulations are pushing fraud teams’ response times, reimbursement policies, and operational procedures.
Cost of fraud management: as the number of transactions, scams, and claims has gone up, so have the operational and technical costs of maintaining a fraud programme. Keeping costs low is a huge focus for FIs, and increased cloud technology and case management tool adoption enable them to address costs head-on.
Identity fraud: AI has made it easier for fraudsters to deepfake documents and make it impossible to verify their authenticity using the human eye. Fraudsters use these documents to bypass identity verification controls and open mule accounts. They also impersonate legitimate individuals to change account information as a part of ATO and conduct fraudulent transactions.
AI is irreplaceable when combating these challenges. It allows us to accurately detect and stop frauds that no other technologies on the market can do today.
‘Where a lot of fraud teams go wrong is they use AI only for fraud detection.’
AI can be used by fraud analytics, strategy, and operations teams, alike to improve detection, streamline processes, and strengthen prevention.
For detection, AI models can continually adapt, looking for anomalous behaviour across payment types and channels, and learning from investigation feedback. Innovative models, like long short-term memory (LSTM) models, transaction trajectory models, and graph analytics, allow us to detect complex patterns, such as mule rings.
For strategy/policy, machine learning (ML) can discover and recommend rules for fraud strategy teams to implement based on data patterns. AI can also benchmark fraud prevention performance against peers and generate business recommendations.
For operations, generative AI (GenAI) can summarise fraud investigation details and speed up investigator reviews. It can also generate SAR narratives, eliminating writing time, and only requiring analysts to confirm outputs before filing.
The UK PSR requires mandatory APP fraud reimbursement and sending and receiving institutions to split liability 50-50. Similar legislation is on the table in the EU, with PSR1, and the US with the Protecting Consumers Against Payment Scams Act.
To prepare, FIs should strengthen four main areas of their programme. Firstly, onboarding and application fraud controls: they should improve new account fraud controls to prevent fraudsters from entering their institution. Secondly, money mule-related controls: strengthen controls by building mule-specific detection models and conducting inbound transaction and early account monitoring. Additionally, they should incorporate known mule lists and network analytics to spot mule accounts quickly.
A third focus should be APP fraud controls, using AI models trained specifically for APP fraud detection (vs. ATO) and creating distinct strategies for each scam type, while cultivating specialised teams trained in each fraud typology and its investigative procedures. Lastly, fraud investigations, namely streamlining investigation processes to counteract the increase in claims from scams and first-party fraud, and improving funds recovery from counterparty institutions. Automation and case management tools will be crucial, enabling dynamic workflows, faster investigations, and automatic funds recovery tracking.
AI presents threats and opportunities for fraud prevention teams. FIs must secure defences across the customer lifecycle, from account opening and early account monitoring to ongoing, cross-payment, and channel fraud monitoring. For optimal fraud prevention, AI must be used pervasively – across fraud detection, strategy, and operations.
NICE Actimize is the largest and broadest provider of fraud, financial crime, risk, and compliance solutions for regional and global financial institutions. Consistently ranked as number one, NICE Actimize experts apply innovative technology to protect institutions and safeguard consumers’ and investors’ assets by identifying financial crime, preventing fraud, and providing regulatory compliance. The company provides real-time, cross-channel fraud prevention, anti-money laundering detection, and trading surveillance solutions. Find us at www.niceactimize.com
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now