AWS Builder Center

Building ScamTrace: An AI-Powered Scam Detection Platform on AWS

How we built ScamTrace to analyze suspicious messages, explain scam patterns, and help users turn a confusing message into actionable next steps using AWS services.

Building ScamTrace: Turning Suspicious Messages Into Actionable Intelligence

Every day, people receive SMS and WhatsApp messages claiming that their bank account will be blocked, their KYC needs to be updated, a parcel requires a payment, or they have won a prize.
The difficult part isn't always recognizing that something feels suspicious.
The difficult part is knowing why it is suspicious and what to do next.
That was the problem we wanted to address with ScamTrace.
ScamTrace is an AI-powered scam awareness platform that allows a user to paste a suspicious message and receive an explanation of the potential scam, the warning signs present in the message, and practical next steps.
The project was built as part of the AWS First Commit hackathon.

The Problem

Scam messages are often designed around urgency and fear.
For example:
"Your bank account will be blocked today. Verify your KYC immediately."
A user may recognize that something is wrong, but several questions immediately follow:
  • Is this actually a scam?
  • What makes the message suspicious?
  • What type of scam is it?
  • What should I do now?
  • Can I report it?
  • Have similar scam patterns appeared before?
We wanted ScamTrace to answer these questions in one place.
Instead of simply returning a binary "scam" or "not a scam" result, the system explains the reasoning in language that a normal user can understand.



What ScamTrace Does

The user starts by pasting a suspicious message into the ScamTrace interface.
The message is sent to our backend, where the AI analyzes it and produces a structured result.
The result includes:
  • Scam verdict
  • Confidence level
  • Scam category
  • Red flags
  • Relevant keywords
  • Explanation
  • Scam mechanism
  • Cybercrime complaint draft
  • Family warning
For example, a KYC phishing message can be identified as a KYC / Banking Phishing attempt.
The system can explain that the message uses urgency and the threat of account blocking to pressure the recipient into taking immediate action.
This makes the result more useful than simply saying:
"This is a scam."



Our AWS Architecture

The project uses a serverless AWS architecture.
React Frontend
|
v
AWS Amplify
|
v
API Gateway
|
v
AWS Lambda
|
+------------------+
| |
v v
Amazon Bedrock Amazon DynamoDB
| |
v v
Amazon Nova 2 Lite Scam Memory
|
v
Structured Analysis
|
v
React Results Dashboard
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article