Pith. sign in

REVIEW 9 cited by

AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2310.17419 v3 pith:KDKXAL36 submitted 2023-10-26 cs.CV

You-Ming Chang , Chen Yeh , Wei-Chen Chiu , Ning Yu This is my paper · ORCID
classification cs.CV
keywords deepfakedetectionmodelsimagefakeaccuracyantifakepromptgenerative
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Deep generative models can create remarkably photorealistic fake images while raising concerns about misinformation and copyright infringement, known as deepfake threats. Deepfake detection technique is developed to distinguish between real and fake images, where the existing methods typically learn classifiers in the image domain or various feature domains. However, the generalizability of deepfake detection against emerging and more advanced generative models remains challenging. In this paper, being inspired by the zero-shot advantages of Vision-Language Models (VLMs), we propose a novel approach called AntifakePrompt, using VLMs (e.g., InstructBLIP) and prompt tuning techniques to improve the deepfake detection accuracy over unseen data. We formulate deepfake detection as a visual question answering problem, and tune soft prompts for InstructBLIP to answer the real/fake information of a query image. We conduct full-spectrum experiments on datasets from a diversity of 3 held-in and 20 held-out generative models, covering modern text-to-image generation, image editing and adversarial image attacks. These testing datasets provide useful benchmarks in the realm of deepfake detection for further research. Moreover, results demonstrate that (1) the deepfake detection accuracy can be significantly and consistently improved (from 71.06% to 92.11%, in average accuracy over unseen domains) using pretrained vision-language models with prompt tuning; (2) our superior performance is at less cost of training data and trainable parameters, resulting in an effective and efficient solution for deepfake detection. Code and models can be found at https://github.com/nctu-eva-lab/AntifakePrompt.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. GlobalForge: Towards Robust AI-Generated Image Detection

    cs.CV 2026-07 conditional novelty 6.0 of 10

    GlobalForge improves AI-generated image detection under real-world degradation by suppressing local shortcuts and enforcing long-range structural reasoning, outperforming prior state-of-the-art by 5.89% average balanc...

  2. AIGI-Holmes: Towards Explainable and Generalizable AI-Generated Image Detection via Multimodal Large Language Models

    cs.CV 2025-07 conditional novelty 6.0 of 10

    AIGI-Holmes combines visual expert pretraining, SFT on explanation data, and direct preference optimization to deliver human-verifiable explanations and top detection accuracy on unseen AI generators.

  3. Interpretable and Reliable Detection of AI-Generated Images via Grounded Reasoning in MLLMs

    cs.CV 2025-06 conditional novelty 6.0 of 10

    Fine-tuning Qwen-2.5-VL on the new FakeXplained dataset of 8,772 AI-generated images with box-and-caption artifact annotations yields an explainable detector with 98.1% accuracy and 37.8% IoU.

  4. AuthGuard: Generalizable Deepfake Detection via Language Guidance

    cs.CV 2025-06 conditional novelty 6.0 of 10

    AuthGuard trains a deepfake vision encoder with MLLM-generated text descriptions plus uncertainty-weighted contrastive learning, improving cross-dataset deepfake detection and adding interpretable LLM reasoning.

  5. Multiverse Through Deepfakes: The MultiFakeVerse Dataset of Person-Centric Visual and Conceptual Manipulations

    cs.MM 2025-06 conditional novelty 5.0 of 10

    MultiFakeVerse provides 845,286 person-centric images edited through VLM-generated instructions; state-of-the-art deepfake detectors and human observers misclassify a large fraction of them.

  6. Simple Domain Generalization for Strong Pixel-Level Image Tampering Detection in Modern VLMs

    cs.CV 2026-07 reject novelty 4.0 of 10

    A simple domain-generalization training recipe (balanced real/tampered batches, late injection of a companion VLM domain, low learning rate) yields large pixel-level tampering-localization gains on out-of-distribution...

  7. Visual Language Models as Zero-Shot Deepfake Detectors

    cs.CV 2025-07 conditional novelty 4.0 of 10

    Zero-shot VLMs scored by normalized yes/no token probabilities beat most trained deepfake detectors on a new SimSwap dataset, and a lightly fine-tuned InstructBLIP is near-perfect on DFDC-P.

  8. LLMs Are Not Yet Ready for Deepfake Image Detection

    cs.CV 2025-06 conditional novelty 4.0 of 10

    On a 100-image benchmark of real and fake faces, ChatGPT, Claude, Gemini, and Grok all fell short of dependable zero-shot deepfake detection, with accuracy varying by category and model.

  9. Unmasking Synthetic Realities in Generative AI: A Comprehensive Review of Adversarially Robust Deepfake Detection Systems

    cs.CR 2025-07 conditional novelty 3.0 of 10

    A systematic review of deepfake detection finds a pervasive lack of adversarial robustness evaluation across all modalities and calls for resilient, modality-agnostic detectors.

Pith tools