Adversative in-context examples ('Sure... However...') defend many LLMs against prefilling jailbreaks but cause over-refusal, so the defense trades safety for helpfulness.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
No Free Lunch for Defending Against Prefilling Attack by In-Context Learning
Adversative in-context examples ('Sure... However...') defend many LLMs against prefilling jailbreaks but cause over-refusal, so the defense trades safety for helpfulness.