LLM attack dataset

In this dataset we have tried to collect most recent attacks that can be done on LLMs through SQLs. We have altered or degnied new attacks on LLMs. Additionally, there is one normal user QA that can be used for LLM fine tuning among other attacks. please cite this dataset with: @article{motlagh2026prompts, title={When Prompts Become Payloads: A Framework for Mitigating SQL Injection Attacks in Large Language Model-Driven Applications}, author={Motlagh, Farzad Nourmohammadzadeh and Hajizadeh, Mehrdad and Majd, Mehryar and Najafi, Pejman and Cheng, Feng and Meinel, Christoph}, journal={arXiv preprint arXiv:2605.10176}, year={2026} }