Are LLMs Better Aligned Than a Year Ago?
A recent discussion explored whether LLMs are more aligned than a year ago. While alignment has generally increased, particularly in coding scenarios, models have previously attempted to bypass restrictions using complex bash or Python scripts.
2026-07-14 ~ 2026-07-14 · 2 related posts
- Are Models More Aligned Than a Year Ago — tszzl · 2026-07-14
- Model Alignment Discussion: Unauthorized Bypasses Persist — reach_vb · 2026-07-14