# LLM-as-a-Discriminator: When Synthetic Tables Still Look Real Source: [LLM-as-a-Discriminator: When Synthetic Tables Still Look Real]() Collection index: [llms.txt]() Data updated: 2026-06-11T05:00:05.512Z Authors: Manel Slokom, Malek Slokom, Thierno Kante Published: 2026-06-01T02:10:02.000Z ## Abstract Privacy and data sharing are often in tension. Many organizations use synthetic data to reduce privacy risk and still share useful data. For tabular data, auditing privacy remains hard. In many cases, even humans cannot easily tell if a table is real or synthetic. In this paper, we propose a method based on LLM discrimination. We ask an LLM to classify each table sample as REAL or SYNTHETIC. We test two settings: C1 with table only, and C2 with table plus distributional metadata. We use LLaMA as an open model and Gemini as a reference model. In our experiments, we run three synthesis models, CTGAN, TVAE, and Gaussian Copula, on two public datasets, UCI Adult and ACS Census. We collect 451 valid trials. Our results show clear differences between models. On Adult, LLaMA reaches DRS=0% in reported cells, while Gemini reaches DRS=100% for CTGAN and TVAE. On Census, LLaMA predicts SYNTHETIC for most samples, while Gemini stays high in C1 but drops for CTGAN and TVAE in C2. We also compare with a classifier two-sample test (C2ST) and record linkage as distributional baselines, and with a human pilot of 2 annotators and 240 trials. Our results show that LLM discrimination is a practical privacy audit signal when model choice, per provider reporting, and data encoding are handled with care. For reproducibility, code and experiment scripts are available at https://github.com/SlokomManel/LLM-as-a-Discriminator. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Disclosure Risk Discrimination on ACS Census](): Disclosure Risk Discrimination; ACS Census; Proportion Real Data: not recorded. - [Synthetic Data Distinguishability on UCI Adult](): Synthetic Data Distinguishability; UCI Adult; p\_hat Real: not recorded. - [Synthetic Data Distinguishability on ACS Census](): Synthetic Data Distinguishability; ACS Census; DRS C1: not recorded. ## Original sources and implementation [Original paper]() [PDF]()