Privacy Peekaboo

Your AI assistant for summarizing Privacy Policies

Comment

Privacy Peekaboo 👻

Inspiration

No one reads privacy policies, so working on a project to summarize how each site handles our data was both a challenging and interesting project.

Overview

PRIVACY_PEEK-A-BOO is an innovative security solution designed to enhance user privacy by evaluating the privacy policies of websites visited through a Chrome extension. This tool aims to provide users with a summarized overview of how their data is being used by different websites, making it easier for them to understand and manage their online privacy.

Features

Chrome Extension

  • PRIVACY_PEEK-A-BOO is implemented as a Chrome extension, making it easily accessible to users who use the Chrome web browser.

Web Scraper for Privacy Policy

  • The extension utilizes a web scraper to extract the privacy policy text from the visited websites automatically.

HTML Element for Report Generation

  • An HTML element is dynamically generated to inform users when a privacy report has been successfully generated for the visited website.

Data Storage and Communication

  • The extracted privacy policy text, along with the URL of the visited website and the user's email, is sent to a server.
  • The server then stores this data in a MongoDB database for future reference and analysis.

Integration with ChatGPT-3.5 API

  • PRIVACY_PEEK-A-BOO communicates with the ChatGPT-3.5 API to generate a summarized overview and advice on how the user's data is being used based on the extracted privacy policy.
  • A special server prompt is used to instruct ChatGPT-3.5 to summarize the web-scraped privacy policy document.

User-Friendly Display

  • The summarized privacy policy information and advice generated by ChatGPT-3.5 are routed back to the visited website to be displayed to the user.
  • A confirmation message is sent back to the Chrome extension to notify the user that the report is ready.

How It Works

  1. Web Scraping: When a user visits a website, the Chrome extension automatically extracts the privacy policy text from the website using a web scraper.

  2. Data Collection: The extracted privacy policy text, along with the URL of the visited website and the user's email, is collected and sent to a server.

  3. Data Storage: The server stores this data in a MongoDB database for future analysis and reference.

  4. Summary Generation: The server communicates with the ChatGPT-3.5 API to generate a summarized overview and advice on how the user's data is being used based on the extracted privacy policy.

  5. Display and Confirmation: The summarized information and advice are routed back to the visited website to be displayed to the user. A confirmation message is sent back to the Chrome extension to notify the user that the report is ready.

Conclusion

PRIVACY_PEEK-A-BOO is a comprehensive security solution that empowers users to take control of their online privacy. By automatically evaluating and summarizing the privacy policies of visited websites, this Chrome extension provides valuable insights into how user data is being used, enabling informed decisions and enhancing user privacy protection online.

Challenges we ran into

The original dataset that we were planning to use was rather unwieldy and hard to manage. It took a long time to be able to parse it correctly and then longer to train our original BERT model. Building a chrome extension was new to all of us so learning the development cycle of a proper extension was both interesting and useful for future projects. We used a MERN fullstack, our server used express.js for endpoints, MongoDB for our database, and held client side HTML in react.

Accomplishments that we're proud of

We are very proud of the Chrome extension working on a good handful of websites, the MERN stack worked and we were able to learn how to manage it, and the success of our ChatGPT-3.5 model.

What we learned

We learned how to set up a Chrome extension, how to set up an express.js server, how to set up a MongoDB database, how to set up a quick website to house all of this, how to train a BERT model, and how to set up a ChatGPT-3.5 api call.

Share this project:

Updates