Watch Your Step: Information Injection in Diffusion Models via Shadow Timestep Embedding

2026-04-30

AI Summary

The security risks and information-hiding potential of timestep embeddings in diffusion models have not been systematically investigated. This work proposes Shadow Timestep Embedding (STE), a novel mechanism that, for the first time, reveals timestep embeddings as an effective side channel capable of carrying covert information. The method enables lossless injection and extraction of hidden data at the scheduler interface. By establishing a theoretical framework based on cross-correlation analysis, the study examines the separability of timestep embeddings and integrates positional encoding modeling with scheduler manipulation techniques. Experiments demonstrate that both adversarial attacks and defensive applications can be successfully implemented without degrading generation quality, thereby opening a new avenue for adversarial generative modeling.

Abstract

Diffusion models have become the foundation of modern generative systems, with most research focusing primarily on improving generation efficiency and output quality. The timestep embedding component is a crucial part of the diffusion pipeline, which provides a temporal conditioning signal to the denoising network, enabling it to adapt its predictions across different noise levels throughout the process. Despite their potential to contain substantial information, timestep embeddings remain underexplored in current research, especially for security risks and reliable provenance. To fill this gap, we introduce Shadow Timestep Embedding (STE), a novel mechanism that investigates the underutilized temporal space for malicious information injection into diffusion models. In particular, when zooming in on the timestep embedding space, we find that different timesteps exhibit distinct representational capabilities that can encode side-channel information. Moreover, such encoded information can be utilized for attack and defense purposes through the scheduler interface. We present a theoretical analysis of timestep embeddings as position-encoding mappings and derive a mutual coherence evaluation that explains the separability of disjoint timestep intervals. Our findings reveal the diffusion model's timestep as a powerful side channel for carrying dedicated information, motivating new directions for adversarial generative modeling by understanding the temporal dimension.

Problem

• diffusion models
• timestep embedding
• information injection
• side-channel
• security risks

Innovation

• Shadow Timestep Embedding
• diffusion models
• side-channel information
• timestep embedding
• adversarial generative modeling