Fun-tuning: Characterizing the Vulnerability of Proprietary LLMs to Optimization-based Prompt Injection Attacks via the Fine-Tuning Interface Leave a Comment / By admin NU / พฤษภาคม 13, 2025