P-block

Inspiration

We were inspired by security for presenters, as well as streamers. At the beginning of the Hackathon, a presenter's API key was exposed, which could have risked a security breach and compromised their system. There wasn't a solution beyond quitting Zoom, which was not possible at the time. Motivated to tackle this common but unsolved problem, we developed a program to actively hide passwords and other potentially security-breaching data, such as phone numbers, bank informaiton and names.

What it does

This program is a privacy-protection system for macOS that monitors the screen in real time to detect and classify confidential information such as API keys, passwords, names, phone numbers, and other personal data, even in multiple languages. When sensitive content is found, it highlights and labels it on the user’s screen while masking it with black bars on the shared display. Users can also input their own personal data (like name, age, or phone number) for the system to recognize and protect. The project is modular, with separate components for detection, frontend display, and screen masking, and includes a README with setup and collaboration instructions for all team members.

How we built it

We utilized Liveshare with VSCode, implemented a llama stack

Challenges we ran into

We faced difficulties ensuring accurate detection across different languages and screen formats, as well as balancing real-time performance with privacy protection without slowing down the system. Since we all had MacOS, it was difficult to bypass or allow security implements to read the screen and identify text/imaging.