AIBearisharXiv – CS AI · Jun 237/10
🧠
The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs
Researchers have discovered that safety mechanisms in large language models operate as linear features in the output layer rather than deep semantic principles, allowing them to be manipulated or inverted through Contrastive Logit Steering. This finding reveals fundamental vulnerabilities in current alignment techniques while simultaneously suggesting a method to strengthen defenses without retraining.
🧠 Llama