We submitted our paper Assessing Per-Sample Membership Inference Vulnerability without Retraining to NeurIPS 2026! Read the preprint here.