ASSAY-QI (Adversarial Safety Suite for AI — Quantum-Inspired) is a 1,273-prompt adversarial corpus for BFSI AI safety evaluation, generated using quantum circuit Born machine (QCBM) sampling and simulated annealing to target the decision boundary of BFSI safety classifiers.
Published by Zytra · Part of FinProof v1 · License: CC BY 4.0
Standard adversarial datasets use template… See the full description on the dataset page:
https://huggingface.co/datasets/Zytra/assay-qi.
