I’m Yusuf Khalid Shire, a Master’s student in the Department of AI Convergence at Kookmin University in Seoul. I work in MCLab with Professor Sang-Chul Kim.
I study how the defenses we put in front of language models fail, and how to measure that failure honestly instead of reporting one flattering number. My paper, PIDS-Bench, does this for prompt-injection detectors: it measures what they do to legitimate security-related text, not only whether they catch attacks. I am also interested in what results like these should mean outside the lab, for safety requirements, deployment standards and AI policy.
What I write here
Most posts are either explanations of attacks and defenses or write-ups of my own research. Both come with the code to rerun them. Research posts end with the experiment I would run next, because that is usually the most useful part.
I write to check my own understanding. If I cannot explain an attack in plain language, with a concrete example and a number I can source, I do not understand it yet.
Publications
- Yusuf Khalid Shire and Sang-Chul Kim. PIDS-Bench: Evaluating Prompt-Injection Detectors Under Over-Defense, Obfuscation, and Distribution Shift. IEEE Access, vol. 14, 2026. DOI · arXiv · benchmark and code
Contact
If you find an error in a post, or rerun my code and get a different number, I want to know: khalidshire@kookmin.ac.kr.