AI RESEARCH
Exploring the Secondary Risks of Large Language Models
arXiv CS.LG
•
ArXi:2506.12382v5 Announce Type: replace Ensuring the safety and alignment of Large Language Models is a significant challenge with their growing integration into critical applications and societal functions. While prior research has primarily focused on jailbreak attacks, less attention has been given to non-adversarial failures that subtly emerge during benign interactions. We