Obscurix
Redacting with Intelligence, Blocking every leak.
Created on 15th June 2025
•
Obscurix
Redacting with Intelligence, Blocking every leak.
The problem Obscurix solves
Obscurix – AI-Powered Data Redaction & Leak Prevention Platform
-
Obsturix tackles the critical issue of data and code leakage by detecting and redacting sensitive information in real time — whether shared as text or image — across platforms like Twitter, Reddit, Stack Overflow, and ChatGPT.
-
The model proactively blocks the sharing of sensitive code or credentials to any AI model (like ChatGPT) or social platforms, while ensuring no data is stored or logged at any stage — making it one of the most secure leak-prevention systems available.
-
With Strict and Creative Modes, Obsturix uses Generative AI to either redact or intelligently rephrase text to preserve tone, reduce negativity, and protect privacy in user communication.
-
Complete with an admin dashboard, OCR-based browser extension to social platforms like Twitter or Reddit, and enterprise-ready safeguards, Obsturix is deployed via AWS EC2 and maintained on GitHub for secure, scalable, and collaborative development.
Challenges we ran into
-
Building models that detect and block sensitive code or text without storing a single byte of user data was a major challenge — requiring custom pipelines that operate in-memory with strict privacy boundaries. But we came over it using GitHub and it pipelines.
-
Ensuring smooth, real-time detection across platforms like ChatGPT, Reddit, Stack Overflow, and Twitter demanded complex integration logic, OCR support for images, and deep pattern recognition to avoid false positives. But it was successfully resolved.
-
We unfortunately were facing issues integrating multiple languages("Spanish", "Hindi" and "Telugu") but still managed to successfully overcome them.
Tracks Applied (3)
GEN AI&ML
Best use of GitHub
GitHub
Best Use of AWS
HackPrix Community