Parallel Hybrid Classical-Quantum Architecture for AI Safety and LLM Jailbreak Detection

Abstract

Large language models (LLMs) are increasingly vulnerable to adversarial "jailbreak" attacks designed to elude safety and privacy controls. Detection is still a challenging task due to the complexity of adversarial prompts and labeled data scarcity. While fine-tuning a traditional Transformer model, such as BERT and its variants, provides strong backbone results with frequently limited recall of complex edge cases, it often comes with extensive computational costs. This work proposes a parallel h

Categories

Framework mappings

Suggested from the entry's categories.

@inproceedings{bajaj2026parallel,
  title = {{Parallel Hybrid Classical-Quantum Architecture for AI Safety and LLM Jailbreak Detection}},
  author = {Sajal Bajaj and Kamlesh Dutta},
  year = {2026},
  month = mar,
  booktitle = {2026 IEEE International Conference on AI Engineering and Innovations (AIEI)},
  doi = {10.1109/AIEI69164.2026.11497307},
  url = {https://www.semanticscholar.org/paper/8afdd6ca08ce0586ebee6b6de85303896c67d376}
}