Back to Projects
AI / Machine Learning

AI Scam Detection

An experimental classifier that flags scam and phishing messages by scoring linguistic and structural risk signals. Paste any message into the demo to see a deterministic, on-device analysis.

INTERACTIVE DEMOPythonscikit-learnNLPFastAPI
View source

Interactive demo

Features

Risk scoring

Combines multiple weak signals into a single 0–100 risk score.

Indicator breakdown

Shows exactly which patterns triggered the score.

Classification

Labels a message as safe, suspicious, or high-risk.

Explainable

Every result comes with a plain-language explanation.

How it works

01

Ingest

The raw message text is normalized and tokenized.

02

Features

Urgency, payment, credential and link signals are extracted.

03

Score

Weighted signals produce a calibrated risk score.

04

Explain

Triggered indicators are surfaced back to the user.

Want to see more? Head back to the full project list.

← Back to Projects