What question did this study set out to answer?

The aim is to evaluate the mechanisms behind pseudo-synthetic data generated by deep learning models.

February 23, 2026Journal of Official Statistics0 citationsOpen Access

Privacy-Enhancing or Privacy-Elusion Technology? A Critical View of (Pseudo)Synthetic Data Based on Deep Learning

Key Points

The aim is to evaluate the mechanisms behind pseudo-synthetic data generated by deep learning models.
Theoretical analysis of deep learning models and pseudo-synthetic data generation.
Discussion of over-parameterization effects in model training.
Margins of safety concerning the use of generated data are articulated.
Identified that pseudo-synthetic data may re-encode training data, raising validity concerns.
Proponents of synthetic data must prove the absence of re-encoding mechanisms in AI models.

Abstract

In this note, we present a plausible structural mechanism by which over-parameterized deep learning models trained on real data may produce pseudo-synthetic data that constitute merely a diﬀerent representation (or re-encoding) of the training data. We conjecture that, in principle, similar mechanisms may be learned by large-scale AI models even if they are not intentionally designed to do so. From there, we derive some cautionary warnings for potential adopters of pseudo-synthetic data generation tools based on deep learning. We claim that the burden of proof that no data re-encoding mechanism is at play in AI-based generation models rests with their proponents.

KI fragen

Bookmark

View Full Paper