Architecture-Agnostic Feature Synergy for Universal Defense Against Heterogeneous Generative Threats

2026-03-16

AI Summary

This work addresses the limitation of existing defenses, which are often confined to specific generative model architectures and thus fail to counter heterogeneous generative threats—resulting in fragmented “defense islands.” To overcome this, we propose ATFS, an architecture-agnostic defense framework that leverages the intrinsic alignment of heterogeneous generative models in high-level feature space. By utilizing target-guided images, ATFS enables cross-architecture gradient coordination without complex calibration. A lightweight feature extractor facilitates efficient adversarial sample optimization under strict pixel perturbation constraints, and the framework supports plug-and-play extension to unseen architectures. Evaluated across diverse generative paradigms—including diffusion models, GANs, and VQ-VAE—ATFS achieves state-of-the-art defense performance (>90% success rate) within 40 iterations and demonstrates strong robustness against common perturbations such as JPEG compression and image resizing.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Agentic search
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Generative AI deployment poses unprecedented challenges to content safety and privacy. However, existing defense mechanisms are often tailored to specific architectures (e.g., Diffusion Models or GANs), creating fragile "defense silos" that fail against heterogeneous generative threats. This paper identifies a fundamental optimization barrier in naive pixel-space ensemble strategies: due to divergent objective functions, pixel-level gradients from heterogeneous generators become statistically orthogonal, causing destructive interference. To overcome this, we observe that despite disparate low-level mechanisms, high-level feature representations of generated content exhibit alignment across architectures. Based on this, we propose the Architecture-Agnostic Targeted Feature Synergy (ATFS) framework. By introducing a target guidance image, ATFS reformulates multi-model defense as a unified feature space alignment task, enabling intrinsic gradient alignment without complex rectification. Extensive experiments show ATFS achieves SOTA protection in heterogeneous scenarios (e.g., Diffusion+GAN). It converges rapidly, reaching over 90% performance within 40 iterations, and maintains strong attack potency even under tight perturbation budgets. The framework seamlessly extends to unseen architectures (e.g., VQ-VAE) by switching the feature extractor, and demonstrates robust resistance to JPEG compression and scaling. Being computationally efficient and lightweight, ATFS offers a viable pathway to dismantle defense silos and enable universal generative security. Code and models are open-sourced for reproducibility.

Problem

• Generative AI
• Universal Defense
• Heterogeneous Threats
• Architecture-Agnostic
• Feature Synergy

Innovation

• Architecture-Agnostic
• Feature Synergy
• Generative Threat Defense
• Gradient Alignment
• Universal Security