Show HN: Privacy Layer for Frontier Models

Category: security

Tags: privacy, llm, data-filtering, pii

Score: 4.7/10 (Innovation: 5, Technical: 4, Documentation: 1, Utility: 5)

A privacy layer intended to filter sensitive information from prompts sent to frontier AI models, presented as a proof of concept. The idea addresses a real concern about leaking personal data to LLMs, but there is very little public information or documentation to evaluate the implementation.

Target audience: end users and developers concerned with PII leakage in LLM prompts

Repository: https://privai-sandy.vercel.app/

View on Hacker News