Authenticator Real Time Sensitive Data Redaction (On Device)

[Aaditesh2307]

🚀 Authenticator: Real-Time Sensitive Data Redaction (On-Device)

Summary

This PR introduces a new functionality that performs on-device, real-time sensitive data redaction before user text is sent to third-party LLMs (ChatGPT, Gemini, etc.).
Sensitive information (PII, financial details, credentials) is detected and masked locally, ensuring that no raw secrets leave the user’s machine.

---

Motivation

• Prevent unintentional leakage of sensitive data to external AI vendors.
• Strengthen compliance with security and privacy regulations.
• Deliver seamless user experience with low-latency, transparent masking.

---

Key Features

• 🔒 On-Device Processing: All detection and masking happens locally.
• 📑 Detection Engine:
  • Regex rules for emails, phone numbers, card numbers, API keys, Aadhaar, SSN, etc.
  • Optional ML-assisted detection for context-based secrets.
• 🛡 Masking Strategies:
  • Emails → te*****@gmail.com
  • Phones → +**-*****-1234
  • Cards → **** **** **** 1234
• ⚡ Performance:
  • <5ms per 1k characters (regex path).
  • <15ms with ML assist on mid-range devices.
• 🎛 Policy Control: Domain-specific rules, allow/deny lists, session-based “hold-to-reveal once”.
• 👁 User Experience:
  • Inline highlight of masked fields.
  • Non-blocking with option to view masked data temporarily.

---