DNA: Uncovering Universal Latent Forgery Knowledge

2026-01-30

AI Summary

This work addresses the challenge that current generative AI produces highly realistic content, rendering traditional detection methods based on superficial artifacts ineffective, while prevailing approaches rely on resource-intensive fine-tuning. To overcome this, we propose the Discriminative Neural Anchors (DNA) framework, which reveals for the first time that pre-trained models inherently encode generalizable forgery detection knowledge in their intermediate layers. DNA activates this latent capability without end-to-end retraining by employing a coarse-to-fine mining mechanism to identify forgery-sensitive discriminative units (FDUs). Integrating feature disentanglement, attention shift detection, triplet fusion scoring, and curvature truncation, DNA achieves superior few-shot performance, demonstrates strong robustness across diverse architectures and unseen generative models, and introduces the high-fidelity synthetic benchmark HIFI-Gen to mitigate data obsolescence.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

As generative AI achieves hyper-realism, superficial artifact detection has become obsolete. While prevailing methods rely on resource-intensive fine-tuning of black-box backbones, we propose that forgery detection capability is already encoded within pre-trained models rather than requiring end-to-end retraining. To elicit this intrinsic capability, we propose the discriminative neural anchors (DNA) framework, which employs a coarse-to-fine excavation mechanism. First, by analyzing feature decoupling and attention distribution shifts, we pinpoint critical intermediate layers where the focus of the model logically transitions from global semantics to local anomalies. Subsequently, we introduce a triadic fusion scoring metric paired with a curvature-truncation strategy to strip away semantic redundancy, precisely isolating the forgery-discriminative units (FDUs) inherently imprinted with sensitivity to forgery traces. Moreover, we introduce HIFI-Gen, a high-fidelity synthetic benchmark built upon the very latest models, to address the lag in existing datasets. Experiments demonstrate that by solely relying on these anchors, DNA achieves superior detection performance even under few-shot conditions. Furthermore, it exhibits remarkable robustness across diverse architectures and against unseen generative models, validating that waking up latent neurons is more effective than extensive fine-tuning.

Problem

• forgery detection
• generative AI
• pre-trained models
• latent knowledge
• hyper-realism

Innovation

• Discriminative Neural Anchors
• Forgery-Discriminative Units
• Feature Decoupling
• Curvature-Truncation Strategy
• HIFI-Gen