Helping the RCMP NC3 Use AI to Process Fraud & Cybercrime Reports
As far too many people in Canada know, fraud and cybercrime cases are on the rise.
The Canadian Anti-Fraud Centre (CAFC) currently receives over 200 reports of fraud and cybercrime a day.
The high number of reports is more than CAFC staff can assess daily, making it difficult to determine larger trends in fraud and cybercrime in Canada.
The RCMP National Cybercrime Coordination Centre (RCMP NC3) wanted to explore the creation of an AI model that could reduce the time it took CAFC staff to assess the reports. They partnered with Code for Canada to see what was possible.
The Challenge
Every day, people in Canada fill out fraud reports. They can do this by calling the CAFC call centre or by visiting its online Fraud Reporting System.
A fraud report includes details of the incident, suspect, and payment method. The CAFC has a database of over 350,000 reports from the past five years alone.
CAFC staff were struggling to scale up triaging and assessing hundreds of reports a day. The RCMP NC3 and CAFC wanted to understand whether an AI model trained on the CAFC's existing database could help improve the form assessment process for over-capacity staff.
The Approach
The C4C team began by conducting interviews with CAFC and RCMP staff focus groups.
Through these conversations, they learned that people filling out fraud reports often select the wrong incident type. CAFC staff must then review the entire report and select the correct incident type, a time-consuming process.
Building on this knowledge, C4C staff:
- Conducted more internal research to better understand the problem
- Held workshops and individual research sessions to measure CAFC staff’s sentiment around working with AI
- Conducted experience mapping to learn where AI could be most beneficial in existing user workflows
Then, C4C staff began working on a prototype to show how an AI model could be created and deployed.
The Results
The C4C team used analytics data and incident statements from the CAFC database to create a list of fraud incident categories.
They then created a simple prototype showing how CAFC staff could interact with an AI model that would automatically suggest the correct incident type, significantly reducing the time it took to assess a form.
The CAFC’s data team took the C4C team’s initial prototype and is now developing its own AI model to improve this important process.
Interested in learning more?
Code for Canada works with governments and non-profits to ethically leverage AI to create better digital services. Interested in learning how we can help? Get in touch today.
