Detects diffusion-generated images 126x faster than reconstruction-based methods by using Gaussian noise disturbance to exploit the statistical 'ease' of fitting synthetic data.
March 17, 2026
Original Paper
FIND: A Simple yet Effective Baseline for Diffusion-Generated Image Detection
arXiv · 2603.14220
The Takeaway
It shifts deepfake detection from expensive, model-dependent reconstruction loops to a simple binary classifier trained on noise-augmented data. This drastically lowers the computational cost of real-time screening while outperforming existing benchmarks.
From the abstract
The remarkable realism of images generated by diffusion models poses critical detection challenges. Current methods utilize reconstruction error as a discriminative feature, exploiting the observation that real images exhibit higher reconstruction errors when processed through diffusion models. However, these approaches require costly reconstruction computations and depend on specific diffusion models, making their performance highly model-dependent. We identify a fundamental difference: real im