Show HN: Abliterated GLM 5.2 model, fine-tuned for security and red-team work

Category: ai-ml

Tags: llm, red-team, security, fine-tuning, content-moderation

Score: 3.7/10 (Innovation: 4, Technical: 4, Documentation: 2, Utility: 3)

Abliteration presents a fine-tuned version of the GLM 5.2 language model intended for security and red-team work, with a policy-based system for allowing, redacting, or rewriting prompts. The project shows a chat demo with rule-based moderation, but lacks a proper README, license, or code repository, making it hard to evaluate or reuse.

Target audience: security researchers, red teams, AI safety engineers

Repository: https://abliteration.ai/#chat

View on Hacker News