SafeNet

• 

  The python code just to show the overview of the project

Inspiration

Millions of elderly and less-educated people are targeted daily by online fraud. Many cannot distinguish between safe and scam messages. I wanted to build a tool that empowers them to stay safe online.

What it does

SafeNet scans text messages (emails, SMS, or chats) and uses algorithms to detect fraud indicators such as suspicious keywords, fake links, and phishing patterns. It classifies messages as Safe ✅ or Suspicious ❌, giving simple warnings for vulnerable users.

How we built it

Python for the core algorithm

Scikit-learn for a lightweight machine learning classifier

NLTK for text preprocessing

Rule-based logic (keyword & URL checks) combined with ML predictions

Prototype tested in a Jupyter/Colab notebook

Challenges we ran into

Limited dataset: scam message datasets are not freely available, so I built a small demo dataset manually.

Balancing simplicity and accuracy: keeping the tool easy to understand while still effective.

Accomplishments that we're proud of

Built a working prototype in a short time.

Designed something meaningful that can genuinely help vulnerable people.

Combined rule-based + ML detection in a lightweight solution.

What we learned

How fraud patterns can be identified through simple text analysis.

Importance of designing for non-technical users — security should be understandable.

The value of combining algorithms with social good.

What's next for SafeNet

Train on a much larger dataset for better accuracy.

Add multi-language support.

Build a browser extension and mobile app for real-time scam detection.

Extend to detect voice scam calls.

Built With

• nltk
• python
• scikit-learn
• streamlit