Showing cs.AIShow all
2 papers · 1 filter
cs.AI2026
From Local Evidence to Safety Verdicts: Causal Tracing in Vision-Language Models
Faezeh Dehghan Tarzjani, Mevan Wijewardena, Alexander Romanus +1
A vision-language model may need to combine an image with a prompt to recognize a safety risk that neither reveals alone. Where does this joint safety judgment become accessible in…
cs.AI2026
Localizing Safety Alignment: MLP Layers and Mid-Network Blocks Encode Refusal Behavior in Large Language Models
Mingyu Zong, Sampad Mohanty, Bhaskar Krishnamachari
Safety alignment in large language models is often treated as a distributed property of the entire network, yet its practical brittleness suggests that refusal behavior may be conc…