Show HN: Viveka: filter LLM output against a Lean-verified Advaita Vedanta model

Category: security

Tags: ai-safety, llm-filter, advaita-vedanta, lean-formalization, python-library

Score: 6.5/10 (Innovation: 7, Technical: 7, Documentation: 7, Utility: 5)

Viveka is a filter layer for LLM outputs that detects objectifying language using a Lean-verified formalization of Advaita Vedanta philosophy, combining machine-verified axioms with heuristic claim extraction to flag or reframe manipulative responses. It is interesting for its novel integration of formal philosophy and AI safety, providing a transparency-focused alternative to typical alignment techniques.

Target audience: backend devs, ai researchers, philosophers

Repository: https://github.com/SpecStudio-net/Viveka · Python · NOASSERTION

View on Hacker News