SecureGO

SecureGO is an all-in-one browser extension that protects users from phishing emails, malicious URLs, and NSFW content — ensuring a safer, smarter internet experience.

Comment

Inspiration

With cyber threats growing every day, we noticed a gap in simple, unified tools that protect users from multiple online dangers at once. Phishing attacks, malicious URLs, and exposure to NSFW content have become common risks — yet most solutions are fragmented or too complex.

We wanted to create a lightweight, fast, and user-friendly extension that empowers everyday users to browse safely and confidently. SecureGO was built to solve a real-world societal need: making internet safety accessible for everyone.

What it does

SecureGO is an all-in-one browser extension that keeps users safe online by:

  • Detecting and warning against phishing emails in real time
  • Identifying and blocking access to malicious or unsafe URLs
  • Blocking NSFW content and warning users before they open inappropriate sites
  • Summarizing user protection stats through a dynamic dashboard
  • Storing detection stats in a database to track protection over time

It runs seamlessly in the background, providing proactive, lightweight, and instant protection without interrupting the browsing experience.

How we built it

  • Backend: Python with Flask for serving models and APIs
  • Frontend: JavaScript for building the browser extension
  • Database: MongoDB for storing detection stats across sessions
  • Phishing email detection: logistic regression ML model trained on two Kaggle datasets, Spam Mails Dataset and Spam Email Dataset
  • Malicious URL detection: random forest classifier ML model trained on 100,000 synthetic URLs
  • Content Filtering: NSFW content detection using the nsfwjs GitHub API
  • Deployment: Local Flask server connected to the browser extension for real-time analysis

Libraries

  • Scikit-learn: for training and deploying the random forest classifier and logistic regression models
  • Pickle & Joblib: for model serialization and efficient storage
  • Pandas & NumPy: for data manipulation, preprocessing, and feature engineering
  • NLTK: for text processing in phishing email detection

Challenges we ran into

  • Lack of Good Malicious URL Dataset: We couldn't find a high-quality, labeled dataset for malicious URLs, so we generated synthetic data (100,000+ records) to train the model effectively.
  • Real-Time Processing: Ensuring that the extension could run in real-time without causing significant delays to the browsing experience was challenging. We optimized by using asynchronous processing for background tasks and minimizing model complexity.
  • NSFW Content Detection: Integrating external APIs like nsfwjs was tricky due to the potential for inaccurate results or performance issues. We had to fine-tune the API calls to balance accuracy and speed, ensuring the user experience was seamless.

Accomplishments that we're proud of

  • Real-Time Protection: Delivered seamless, real-time detection for phishing emails, malicious URLs, and NSFW content, enhancing user security without interrupting their browsing experience.
  • Synthetic Dataset Innovation: Created a synthetic dataset of 100,000 URLs, overcoming the lack of a good malicious URL dataset. This enabled us to build an accurate classifier for URL safety detection.
  • User-Centric Dashboard: Built a dynamic dashboard that gives users a clear view of their protection stats, including phishing emails detected, malicious URLs blocked, and NSFW content filtered, helping them stay informed.
  • Data Persistence: Implemented a database to track detection stats across multiple sessions, allowing users to see how their protection evolves over time.

What we learned

  • Browser Extension Development: Developing a browser extension introduced us to the complexities of building lightweight, cross-browser compatible applications. We learned how to manage browser security restrictions, handle permissions, and ensure consistent functionality across Chrome.
  • Importance of Data Quality: We learned that having high-quality, diverse datasets is crucial for training accurate models. The lack of a suitable malicious URL dataset pushed us to create our own synthetic data, which turned out to be an invaluable learning experience.
  • Model Optimization for Real-Time Use: Training models for real-time applications required us to optimize both the model's performance and speed. We learned how to balance accuracy with efficiency, ensuring minimal impact on the user’s browsing experience.
  • User Experience Matters: Focusing on a lightweight, non-intrusive experience was crucial. The user’s perception of safety is as important as the actual protection, so we prioritized clear alerts, easy-to-understand dashboards, and a smooth overall experience.

What's next for SecureGO

  • Expanding to Other Browsers: We plan to extend SecureGO’s protection to Safari, Edge, and other popular browsers, ensuring more users can benefit from enhanced security.
  • Built-in Secure VPN Integration: Introducing a secure VPN feature to provide an extra layer of privacy and protection while users browse, ensuring their data remains safe from threats like hackers and trackers.
  • SecureGO API: Developing a SecureGO API to allow third-party apps and services to integrate phishing, URL safety, and NSFW content detection into their platforms, broadening the reach of our security technology.
Share this project:

Updates