Auditing Privacy Leakage in Tabular Foundation Model Embeddings

Xun Wang, Adam Dziedzic, Michael Backes, Franziska Boenisch. NeurIPS 2026.

We study how much sensitive information can be recovered from tabular foundation-model embeddings. Our audit examines attribute recovery from embeddings alone and how additional side information changes privacy risk.

Recommended citation: Wang, X., Dziedzic, A., Backes, M., & Boenisch, F. (2026). "Auditing Privacy Leakage in Tabular Foundation Model Embeddings." Conference on Neural Information Processing Systems (NeurIPS).