AI RESEARCH
False Friends in the Shell: Unveiling the Emoticon Semantic Confusion in Large Language Models
arXiv CS.AI
•
ArXi:2601.07885v2 Announce Type: replace-cross Emoticons are widely used in digital communication to convey affective intent, yet their safety implications for Large Language Models (LLMs) remain largely unexplored. In this paper, we identify emoticon semantic confusion, a vulnerability where LLMs misinterpret ASCII-based emoticons to perform unintended and even destructive actions.