Safety Alignment in NLP Tasks: Weakly Aligned Summarization as an In-Context Attack.

Published in: CoRR (2023)

Keyphrases