# Prefill-Guided Thinking for zero-shot detection of AI-generated images

arXiv: https://arxiv.org/abs/2506.11031 · HTML: https://mlsanity.com/paper/2506.11031/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 31 of its 38 references that are in the map.
- First posted on arXiv 20 May 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 30 of 31)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Diffusion Models Beat GANs on Image Synthesis](https://mlsanity.com/paper/2105.05233/) (2021)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://mlsanity.com/paper/2307.01952/) (2023)
- [GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models](https://mlsanity.com/paper/2112.10741/) (2021)
- [Large Scale GAN Training for High Fidelity Natural Image Synthesis](https://mlsanity.com/paper/1809.11096/) (2018)
- [Xception: Deep Learning with Depthwise Separable Convolutions](https://mlsanity.com/paper/1610.02357/) (2016)
- [FaceForensics++: Learning to Detect Manipulated Facial Images](https://mlsanity.com/paper/1901.08971/) (2019)
- [StarGAN v2: Diverse Image Synthesis for Multiple Domains](https://mlsanity.com/paper/1912.01865/) (2019)
- [StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery](https://mlsanity.com/paper/2103.17249/) (2021)
- [CNN-generated images are surprisingly easy to spot... for now](https://mlsanity.com/paper/1912.11035/) (2019)
- [Leveraging Frequency Analysis for Deep Fake Image Recognition](https://mlsanity.com/paper/2003.08685/) (2020)
- [Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models](https://mlsanity.com/paper/2212.03860/) (2022)
- [MesoNet: a Compact Facial Video Forgery Detection Network](https://mlsanity.com/paper/1809.00888/) (2018)
- [DeepfakeBench: A Comprehensive Benchmark of Deepfake Detection](https://mlsanity.com/paper/2307.01426/) (2023)
- [Raising the Bar of AI-generated Image Detection with CLIP](https://mlsanity.com/paper/2312.00195/) (2023)
- [Learning Rich Features for Image Manipulation Detection](https://mlsanity.com/paper/1805.04953/) (2018)
- [Fourier Spectrum Discrepancies in Deep Network Generated Images](https://mlsanity.com/paper/1911.06465/) (2019)
- [DF40: Toward Next-Generation Deepfake Detection](https://mlsanity.com/paper/2406.13495/) (2024)
- [Collaborative Diffusion for Multi-Modal Face Generation and Editing](https://mlsanity.com/paper/2304.10530/) (2023)
- [Detecting and Grounding Multi-Modal Media Manipulation](https://mlsanity.com/paper/2304.02556/) (2023)
- [Can ChatGPT Detect DeepFakes? A Study of Using Multimodal Large Language Models for Media Forensics](https://mlsanity.com/paper/2403.14077/) (2024)
- [FreqBlender: Enhancing DeepFake Detection by Blending Frequency Knowledge](https://mlsanity.com/paper/2404.13872/) (2024)
- [Perceptual Artifacts Localization for Image Synthesis Tasks](https://mlsanity.com/paper/2310.05590/) (2023)
- [DiffusionFake: Enhancing Generalization in Deepfake Detection via Guided Stable Diffusion](https://mlsanity.com/paper/2410.04372/) (2024)
- [Contrasting Deepfakes Diffusion via Contrastive Learning and Global-Local Similarities](https://mlsanity.com/paper/2407.20337/) (2024)
- [Characteristics and prevalence of fake social media profiles with AI-generated faces](https://mlsanity.com/paper/2401.02627/) (2024)

## Abstract (arXiv)

Traditional supervised methods for detecting AI-generated images depend on large, curated datasets for training and fail to generalize to novel, out-of-domain image generators. As an alternative, we explore pre-trained Vision-Language Models (VLMs) for zero-shot detection of AI-generated images. We evaluate VLM performance on three diverse benchmarks encompassing synthetic images of human faces, objects, and animals produced by 16 different state-of-the-art image generators. While off-the-shelf VLMs perform poorly on these datasets, we find that prefilling responses effectively guides their reasoning -- a method we call Prefill-Guided Thinking (PGT). In particular, prefilling a VLM response with the phrase "Let's examine the style and the synthesis artifacts" improves the Macro F1 scores of three widely used open-source VLMs by up to 24%. We analyze this improvement in detection by tracking answer confidence during response generation. For some models, prefills counteract early overconfidence -- akin to mitigating the Dunning-Kruger effect -- leading to better detection performance.