Observation-Level Watermarking and Detection for Tabular Data

2026-07-12

AI Summary

Existing watermarking methods struggle to effectively handle discrete, categorical, and mixed-type tabular data and lack reliable single-sample detection mechanisms. This work proposes STAMP, a novel framework that, for the first time, enables observation-level watermark embedding and single-sample detection for mixed-type tabular data. By leveraging a distribution-preserving strategy for watermark insertion and integrating a statistical detection mechanism, STAMP achieves high detection accuracy and provides theoretical guarantees of asymptotic consistency while maintaining strong fidelity to the original data distribution. Extensive simulations and evaluations on two real-world datasets demonstrate the method’s effectiveness, robustness, and superior ability to preserve the underlying data distribution.

Abstract

With the development of generative AI, watermarking techniques have been widely used to detect the authenticity of AI-generated data and protect the rights of users and creators. While it is already well applied in data types including imaging and text data, watermarking tabular data is still under-explored. Existing methods primarily focus on numerical data, leaving discrete, categorical, and mixed data less studied. In this work, we propose STAMP (Single-observation Tabular Attribution and Marking Procedure), a novel framework for watermarking tabular data that can accommodate and preserve a wide range of distributions. We also develop a corresponding detection mechanism, which can reliably identify watermarks even when the sample size is as small as one. We establish theoretical guarantees for asymptotic consistency and detection accuracy. Finally, through extensive simulation studies and two real-data applications, we demonstrate that the proposed method is effective and robust to subsetting, while maintaining data fidelity and a high detection rate.

Problem

• tabular data
• watermarking
• categorical data
• single-observation detection
• data authenticity

Innovation

• tabular data watermarking
• single-observation detection
• STAMP
• generative AI attribution
• data fidelity