ML Security PapersExplainable Detection of AI-Generated Images with Artifact Localization Using Faster-Than-Lies and Vision-Language Models for Edge Devices
Aryan Mathur , Asaduddin Ahmed , Pushti Amit Vasoya , Simeon Kandan Sonar , Yasir Z , Madesh Kuppusamy
Published on arXiv
2510.23775
Output Integrity Attack
OWASP ML Top 10 — ML09
Key Finding
Achieves 96.5% accuracy on adversarially-augmented CiFAKE dataset with 175ms inference time on 8-core CPUs, enabling edge deployment
Faster-Than-Lies
Novel technique introduced
The increasing realism of AI-generated imagery poses challenges for verifying visual authenticity. We present an explainable image authenticity detection system that combines a lightweight convolutional classifier ("Faster-Than-Lies") with a Vision-Language Model (Qwen2-VL-7B) to classify, localize, and explain artifacts in 32x32 images. Our model achieves 96.5% accuracy on the extended CiFAKE dataset augmented with adversarial perturbations and maintains an inference time of 175ms on 8-core CPUs, enabling deployment on local or edge devices. Using autoencoder-based reconstruction error maps, we generate artifact localization heatmaps, which enhance interpretability for both humans and the VLM. We further categorize 70 visual artifact types into eight semantic groups and demonstrate explainable text generation for each detected anomaly. This work highlights the feasibility of combining visual and linguistic reasoning for interpretable authenticity detection in low-resolution imagery and outlines potential cross-domain applications in forensics, industrial inspection, and social media moderation.
Key Contributions
- Lightweight 'Faster-Than-Lies' CNN classifier achieving 96.5% accuracy on CiFAKE with 175ms inference on 8-core CPUs for edge deployment
- Autoencoder-based reconstruction error heatmaps for localizing visual artifacts in 32×32 AI-generated images
- Integration with Qwen2-VL-7B to generate natural-language explanations for 70 categorized artifact types across eight semantic groups
🛡️ Threat Analysis
Primary contribution is a novel AI-generated image detection system — classifying real vs. synthetic images, localizing artifacts, and explaining decisions — which falls squarely under output integrity and AI-generated content detection.