Counter Fraud Strategy in the Generative AI Era

At Counter Fraud Conference 2024, Amazon Web Services (AWS) hosted a discussion on ‘Real-time Fraud Detection Using AI, Generative AI and Graph Neural Networks’ with panellists from the Financial Conduct Authority (FCA) and Department for Work and Pensions (DWP). The discussion was followed by wide-ranging conversations with investigators, civil servants, and data professionals supporting the government’s fraud prevention initiatives across the public sector. In this article, Deepak Shukla, Head of Data and AI for the public sector at AWS provides his reflections on the discussion at the conference.

AWS Fraud Blog Panellist Image

Data and AI are taking centre stage in the government’s counter-fraud strategy. Public sector customers are looking to maximise the value of their cloud platforms as they move towards real-time fraud prevention. Below are a few themes shared by customers in various interactions at the conference.

  1. Organisations want to focus more on real-time proactive fraud prevention measures. The operational cost of recovery is becoming too high.
  2. Adding fraud prevention measured within citizens’ digital journey will impact frictionless experiences.
  3. The public sector needs an integrated approach to fraud investigations, reducing dependencies on data scientists/engineers during an investigation.
  4. Current data and analytics solutions have very high licensing costs and need to explore more cloud-native services to optimise cost.
  5. Lots of work is in progress on regulating data sharing across agencies for fraud investigation, but the process is manual with bespoke spreadsheets, and data files created exposing governance.

Advancements in Generative Artificial Intelligence (Generative AI) offer exciting opportunities for public sector organisations to automate systems and controls to protect themselves and citizens from financial crime and risk. There is a potential for fraudsters deploying Generative AI to commit different frauds. There are cases where bad actors are already using AI for synthetic identity fraud, document forgery fraud, AI-enabled identity fraud and deepfakes & voice spoofing, and this list will continue to grow with time. So, Generative AI technology is bound to play a leading role in the way the public sector fights AI-led fraud.

Beyond technology, real-time fraud prevention requires a mindset shift in the way organisations approach fraud solutions. This needs to be looked at similarly to how we design and shape citizen’s digital journeys. The process starts with assessing citizen’s fraud exposure across their end-to-end journey and touchpoints with government agencies (Acquire, Onboard, Service and Maintain). Identify fraud types at each stage of the citizen’s journey and embed prevention checks and solutions. Use data to gain a more holistic understanding of the citizen and context. Create and manage fraud risk score based on customer’s historical interactions with the agency and other third parties. And assist customers via their preferred channels about ongoing fraud exposureguiding customers towards personalized fraud prevention educationFuture fraud solutions require easy collaboration between investigators, citizens, and data engineers/scientists.

As organisations build robust fraud prevention capabilities, here are four key foundational components they need to focus on:

  1. Generative AI-powered process automation–Use managed AI services to automate citizen journeys for claims and public agency touchpoints. To navigate the skills gap challenge and faster time to value adopt managed AI services on cloud for image recognition, identity management, document processing and fraud detection.
  2. Data Foundations–Strong data integration, preparation and storage, along with metadata management and cataloguing, are essential. Advanced entity resolution and network graph analytics supported by natural language processing and automated data sharing and governance capabilities.
  3. Investigation and Self-Service–The insights and fraud scores should be able to integrate with any case management solution to drive insights-driven investigations. The systems should support no-code business personas with reduced dependencies on data scientists/engineers. Generative AI-powered assistants will play a key part in navigating the skills gaps.
  4. Consumption and Notification–Establish a common smart call centre integration with a chatbot and the ability to send real-time alerts and notifications to citizens. Engage citizens in real-time to proactively prevent them from fraud and educate them on new types of fraud.

As you shape your counter-fraud strategy and speed up your real-time fraud prevention journey, AWS has tailored programmes you could leverage. Data Driven Everything (D2E) programme is a proven method supporting the broader transformation required to detect and prevent fraud. AWS’s Generative AI Innovation Centre helps with ideation and rapid prototyping to speed up your journey to outcomes. Please get in touch to discuss more by sharing your details through this link.