Static refusal tests overstate the safety of skill-merged LLMs: models with identical static safety differ sharply under adaptive attack, and a task-vector overlap with a safety subspace flags only same-recipe abliteration-style erosion.
SafeMERGE head-to-head (main paper§7).We compare against the selective-layer baseline Safe- MERGE (Djuhera et al
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
When Skills Meet Safety: Benchmarking and Characterizing the Adaptive Jailbreak Robustness of Skill-Merged LLMs
Static refusal tests overstate the safety of skill-merged LLMs: models with identical static safety differ sharply under adaptive attack, and a task-vector overlap with a safety subspace flags only same-recipe abliteration-style erosion.