mlsanityData to 14 Oct 2026

Understanding the Impact of Data Domain Extraction on Synthetic Data Privacy

arXiv 2504.08254 ↗First posted 11 Apr 2025cs.CR cs.LGcited by 0

On the map: LLM safety and security → Membership inference attacks

Abstract (arXiv)

Privacy attacks, particularly membership inference attacks (MIAs), are widely used to assess the privacy of generative models for tabular synthetic data, including those with Differential Privacy (DP) guarantees. These attacks often exploit outliers, which are especially vulnerable due to their position at the boundaries of the data domain (e.g., at the minimum and maximum values). However, the role of data domain extraction in generative models and its impact on privacy attacks have been overlooked. In this paper, we examine three strategies for defining the data domain: assuming it is externally provided (ideally from public data), extracting it directly from the input data, and extracting it with DP mechanisms. While common in popular implementations and libraries, we show that the second approach breaks end-to-end DP guarantees and leaves models vulnerable. While using a provided domain (if representative) is preferable, extracting it with DP can also defend against popular MIAs, even at high privacy budgets.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Membership inference attacks, one of 12 directions in LLM safety and security. Builds on 14 of its 35 references that are in the map. First posted on arXiv 11 Apr 2025.

Builds on

14 of 35 references.

References · 14

  1. 2021Winning the NIST Contest: A scalable and general approach to differentially private synthetic datacited by 102
  2. 2019Graphical-model based estimation and inference for differential privacycited by 74
  3. 2019Diffprivlib: The IBM Differential Privacy Librarycited by 53
  4. 2020New Oracle-Efficient Algorithms for Private Synthetic Data Releasecited by 43
  5. 2023SoK: Privacy-Preserving Data Synthesiscited by 38
  6. 2023Achilles' Heels: Vulnerable Record Identification in Synthetic Data Publishingcited by 28
  7. 2015Differentially Private k-Means Clusteringcited by 27
  8. 2016PrivTree: A Differentially Private Algorithm for Hierarchical Decompositionscited by 22
  9. 2024"What do you want from theory alone?" Experimenting with Tight Auditing of Differentially Private Synthetic Data Generationcited by 21
  10. 2024Differentially Private Release of Israel's National Registry of Live Birthscited by 17
  11. 2024The Elusive Pursuit of Reproducing PATE-GAN: Benchmarking, Auditing, Debuggingcited by 13
  12. 2022dpart: Differentially Private Autoregressive Tabular, a General Framework for Synthetic Data Generationcited by 12
2 more
  1. 2023Synthetic Data: Methods, Use Cases, and Riskscited by 10
  2. 2025The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Datacited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Synth-MIA: A Testbed for Auditing Privacy Leakage in Tabular Data Synthesis
  2. 2025Ensembling Membership Inference Attacks Against Tabular Generative Models
  3. 2024SafeSynthDP: Leveraging Large Language Models for Privacy-Preserving Synthetic Data Generation Using Differential Privacy
  4. 2023Synthetic is all you need: removing the auxiliary data assumption for membership inference attacks against synthetic data
  5. 2025Privacy Auditing Synthetic Data Release through Local Likelihood Attacks
  6. 2025The DCR Delusion: Measuring the Privacy Risk of Synthetic Data
  7. 2025Membership Inference over Diffusion-models-based Synthetic Tabular Data
  8. 2025When Privacy Isn't Synthetic: Hidden Data Leakage in Generative AI Models