Pith. sign in

REVIEW 4 major objections 5 minor 60 references

Evidence-Grounded Forensic Reasoning for Detecting and Grounding Multi-Modal Media Manipulation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper shows that a multimodal manipulation detector can make its reasoning auditable by anchoring every cited piece of evidence to the predicted face box or text span, and then verifying that match with a rule-based reward system.

desk verdict Real gains on detection and a genuinely new token-level routing mechanism, but the paper's headline claim about evidence–anchor consistency is enforced as a training reward and never actually measured at test time. read the letter →

arxiv 2608.08009 v1 pith:XQJSXLCQ submitted 2026-08-08 cs.CV cs.AI

Yichun Yeh , Yiheng Li , Xiaobo Hu , Zhen Lei , Yang Yang This is my paper · ORCID
classification cs.CVcs.AI
keywords mediamanipulationdetectionmulti-modalforensicsgroundedreasoninganchor-and-verifymultimodallargelanguagemodelverifiablerewardtoken-leveladvantageroutingDGM4
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to make multi-modal manipulation detection auditable: instead of a black-box fake/real label, a detector should explain its verdict by pointing at the precise face region or text token that was altered, and every sentence of the explanation should refer back to that same location. On the DGM4 benchmark, which combines authenticity classification, manipulation-type identification, and face-box and text-token grounding, the authors claim their Evidence-Grounded Forensic Reasoning (EFR) framework reaches the best reported detection accuracy and competitive grounding while producing reasoning records whose cited evidence is spatially consistent with the predicted anchors. This matters for forensic practice because fact-checkers and legal reviewers need to challenge a decision by inspecting its evidence; a rationale that points elsewhere is as unreliable as no rationale at all.

What carries the argument

The load-bearing object is the Anchor-and-Verify reasoning chain, a structured four-block output format together with the reward that enforces it. The Forensic Hypothesis Block places the predicted coordinates at the start of the sequence, making the face box $b$ and text span $s$ explicit anchors, and the Physical Evidence Binding block requires downstream citations to reference locations matching those anchors. MDA routing then uses the XML block boundaries to assign per-task advantages to per-task token positions, so gradients from the classification head do not wash out gradients from the face-box or text-span heads, and gradients from one localization head do not interfere with the other.

What would settle it

A reader could audit the evidence-binding blocks of a large sample of correctly predicted DGM4 images and count how often a cited box fails to overlap the declared anchor by IoU below 0.5, or a cited token span falls outside the declared span; a non-negligible mismatch rate would falsify the central claim of coordinate-level consistency.

Watch

Extended reading notes

Core claim

The central claim is that the grounding gap in MLLM-based forgery detection can be closed by making the model's conclusion coordinates the anchor for all downstream reasoning. EFR outputs a structured chain: a forensic hypothesis block containing the predicted verdict, fine-grained manipulation type, face bounding box, and text-token span, followed by modality-isolated perception, cross-modal conflict analysis, and physical evidence binding in which every visual citation must satisfy $b' \approx b$ and every text citation must satisfy $s' \subseteq s$. Consistency is enforced rather than hoped for: a five-component rule-based reward system scores format, classification, face localization, text localization, and anchor-evidence consistency, while the Modality-Decoupled Advantage (MDA) mechanism routes each task-specific advantage only to the tokens that produced that output. Trained first by supervised fine-tuning on 50K curated reasoning chains and then by GRPO with token-level routed advantages, EFR reports the best binary detection and multi-label F1 on DGM4 and grounding close to the strongest dedicated heads.

Load-bearing premise

The load-bearing premise is that the 50,000 example reasoning chains used for training are reliable enough to teach genuinely grounded evidence, rather than a plausible-sounding imitation of grounding.

Editorial extensions

If this is right

  • DGM4-style tasks can carry an auditable record: every conclusion comes with a checkable location that a human or another algorithm can inspect.
  • Anchor-evidence consistency can be measured automatically, so reasoning quality is not judged only by fluency or template overlap.
  • The two-stage recipe of cold-start supervised fine-tuning followed by reward optimization with token-level advantage routing may transfer to other multi-head multimodal generation tasks.
  • Stage 2 optimization can rebalance an asymmetric operating point, trading precision for recall where supervised training provides no signal for partial span recovery.
  • The reported gains are concentrated in fine-grained attribution rather than binary detection, suggesting that structured reasoning helps most when a model must distinguish among manipulation types.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial extension: the coordinate-consistency check could be applied at inference time to any multimodal explanation, turning how well cited boxes and spans match predicted anchors into a post-hoc trust score.
  • Editorial extension: because the reward system is deterministic and rule-based, the pipeline could be adapted to other structured output tasks, such as document-level fact-checking or medical image reporting, wherever outputs can be parsed into typed blocks.
  • Editorial extension: if the 50K reasoning dataset inherits perceptual habits from the annotation model, cross-dataset evaluation on newer or harder forgery benchmarks would test whether the grounding transfers; the paper does not run that test.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes EFR, a two-stage framework for the DGM4 task that combines a curated 50K forensic reasoning dataset, an Anchor-and-Verify structured chain-of-thought output format, and a GRPO-based reinforcement-learning stage with five rule-based rewards and a Modality-Decoupled Advantage (MDA) routing mechanism. The authors claim state-of-the-art detection, competitive grounding, and, centrally, that the generated reasoning records bind cited evidence to the predicted conclusion coordinates at the spatial level.

Significance. If the evidence-binding property were measured and shown to hold, the paper would make a useful contribution to explainable media forensics, where auditable reasoning is a practical requirement. The framework is clearly motivated, the proposed decomposition of the training signal into classification, localization, and consistency rewards is principled, and the reported gains on CF1/OF1 over strong baselines are notable. The ablation study is also careful in isolating the effects of MDA components. However, the central claim of anchor-evidence consistency is never directly evaluated as a test-time property, and several experimental-protocol and reproducibility issues currently prevent full assessment of the state-of-the-art claim.

major comments (4)
  1. [Section 5 / Section 4.4] The paper's central claim, that EFR produces 'reasoning records whose cited evidence is spatially consistent with the predicted anchors,' is never evaluated as a held-out property. Table 4 reports NLI Coherence, Redundancy, ROUGE-L, and Distinct-2/3 on the <evidence_binding> and <perception> blocks; these are text-level metrics and do not check whether the box/span coordinates cited in the evidence actually match the coordinates in the <conclusion> anchors. Because R_consistency (Eq. 6) is an explicit training reward, any improvement in anchor-evidence agreement would be a direct consequence of the optimization target, not an emergent discovery. Please add a quantitative test-time evaluation that parses the evidence coordinates, compares them with the conclusion anchors (e.g., IoU for boxes and token-overlap for spans), and reports the agreement rate on the test set, ideally also on an out-of-distribution set.
  2. [Section 3.2.2 / Figure 2] The 50K forensic reasoning dataset is generated by conditioning Qwen3-VL-32B on ground-truth findings and spatial priors, and Stage-1 SFT trains the model to reproduce these reasoning chains. This means the evidence cited in the training data already matches the ground-truth anchors by construction, so R_consistency mainly reinforces what the cold-start model has already learned from the annotation data. The reported consistency is therefore confounded with the training-data construction. To support the generalization claim that EFR learns to bind evidence to anchors, the authors should evaluate anchor-evidence consistency on samples whose ground-truth anchors were not used to generate training reasoning, or provide an ablation that removes the ground-truth conditioning from the data-generation pipeline.
  3. [Table 1 / Section 4.1] The mAP comparison is not on equal footing across methods. For MLLM-based methods, mAP is computed from token log probabilities or from hard 0/1 predictions, while discriminative baselines report mAP from continuous multi-label scores. The paper acknowledges this but still presents EFR's mAP (90.41/81.58) in the same table and describes it as 'competitive.' Under the log-probability protocol EFR is actually below MSF (91.42) and IDseq (90.01), and the hard-0/1 value is not directly comparable to any baseline. To support a state-of-the-art claim, please report mAP under the identical protocol used for prior work, or provide statistical evidence that the protocol difference does not affect the ranking of methods.
  4. [Tables 1-3 / Section 4.4] No error bars, confidence intervals, or significance tests are reported, and several key differences are small: in Table 2, ACC changes from 90.74 to 90.82 and mAP from 78.88 to 81.58 with substantial intermediate variation; in Table 3, mAP varies by 0.01-0.1 across several configurations. Without variance estimates, it is unclear whether the reported improvements are meaningful. In addition, the reasoning-quality evaluation in Table 4 is restricted to 'correctly predicted manipulated samples,' which conditions on the model's own correctness and may inflate reasoning metrics by selecting easy cases. Please report results across multiple seeds and evaluate reasoning metrics on the full test set or with a matched-control design.
minor comments (5)
  1. [Abstract] There is a typo: 'mitigats' should be 'mitigates.'
  2. [Section 3.3.2 / Eq. (8)] The definition of the token-position sets T_k is not formalized, and the statement that R_format 'scales the whole-sequence reward' is not reflected in Eq. (8); please specify how the format gate enters the token-level objective.
  3. [Section 3.3.1] The text says 'The full decomposition is given in the supplementary,' but no supplementary material is included in this version; please include it or provide a link.
  4. [Section 4.4 / Figure 4] The qualitative example is illustrative but does not substantiate the coordinate-level consistency claim on its own; please supplement it with the quantitative metric requested in Major Comment 1.
  5. [General] The paper does not state whether the 50K reasoning dataset, the code, or the trained models will be released; for reproducibility, please add an availability statement.

Circularity Check

1 steps flagged · score 6.0 of 10

The explainability claim—cited evidence is spatially consistent with predicted anchors—is the exact property encoded in the R_consistency reward, and no held-out coordinate-binding metric is reported; detection and grounding results remain externally benchmarked.

  1. self definitional [Section 3.3.1 (Eq. 6) and Section 5 Conclusion]
    "Consistency (R_consistency) verifies internal coherence of the reasoning chain, rewarding cases where the evidence cited in the evidence_binding block spatially and semantically matches the conclusion anchors ... R = λf R_format + λc R_cls + λv R_face + λt R_text + λr R_consistency ... producing reasoning records whose cited evidence is spatially consistent with the predicted anchors."

    The conclusion's distinguishing qualitative claim is exactly the quantity R_consistency is defined to reward, and Eq. (6) makes that reward part of the training objective. Any policy that maximizes the reward is, by construction, pushed toward evidence that matches its own conclusion anchors. No held-out metric in Section 4 measures coordinate-level evidence-anchor binding: Table 4 scores NLI coherence, redundancy, ROUGE-L, and Distinct-2/3, all text-level, and Table 3 shows only mAP/IoU/F1 changes within 0.1 points when R_consist is added. Additionally, the SFT cold-start targets were generated by conditioning Qwen3-VL-32B on ground-truth findings and spatial priors (Section 3.2.2), so evidence-anchor agreement was already present in the training inputs.

full rationale

The DGM4 detection and grounding results are not circular: EFR is benchmarked against external methods (HAMMER, IDseq, ASAP, etc.) on standard metrics, and those numbers stand independently of the paper's self-citations. The self-citations that appear (CSCL [18], VERITAS [40]) are related-work summaries, not load-bearing premises. The circular component is confined to the explainability contribution. The Anchor-and-Verify chain is defined as requiring downstream evidence to match conclusion coordinates; the same definition is hard-coded into the R_consistency reward in Eq. (6). The conclusion asserts, as an experimental outcome, that EFR produces reasoning records whose cited evidence is spatially consistent with the predicted anchors, but the only evidence offered is a single qualitative example (Figure 4) and text-level metrics that do not measure coordinate binding. That asserted outcome is entailed by the training objective, so it is not an independent empirical discovery. Because the central detection and grounding claims retain independent content, the overall circularity is partial rather than total.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The framework relies on a substantial set of hand-chosen reward weights, scoring curves, and dataset construction choices that are not released or fully specified. The most consequential axiom is that DGM4 ground truth is accurate and that the MLLM-generated reasoning text, after filtering, is a valid teacher for forensic evidence. The reward system also assumes that spatial overlap between cited boxes/spans and predicted anchors is a sufficient definition of 'grounded evidence.'

free parameters (5)
  • Reward weights lambda_f, lambda_c, lambda_v, lambda_t, lambda_r = not reported
    Hand-chosen weights in Eq. (6) that balance format, classification, face localization, text localization, and consistency rewards; no sensitivity analysis is reported.
  • Hierarchical classification reward weights w_fam, w_sub, w_exact = not reported
    Fixed weights in Eq. (5) for family-level, subtype-level, and exact-match rewards; the paper states they are fixed but does not give their values or a tuning procedure.
  • IoU tiers and Jaccard curves for R_face and R_text = not reported
    The mapping from box IoU to tiered reward and the subtype-specific text overlap curves are defined in the supplementary; these hand-designed scoring functions control the localization reward.
  • Task-aware resampling thresholds = not reported
    The criteria for exempting groups and the intensity of extra sampling are described qualitatively in Sec. 3.3.2; the exact thresholds are not specified.
  • Data stratification and sample sizes = 147K/53K/7K tiers; 80K annotated; 50K final; 10K RL
    The tier cutoffs and the choice to annotate 80K and fine-tune on 50K/10K are arbitrary design decisions that shape the model; no justification is given for these specific numbers.
assumptions (4)
  • standard math GRPO objective (Eq. 2) provides valid gradient estimates for policy optimization.
    The paper inherits GRPO from DeepSeekMath [36] and uses it as the RL backbone; no new proof is provided.
  • domain assumption DGM4 ground-truth labels (face boxes, text spans, manipulation types) are accurate enough for supervised training and evaluation.
    The entire data construction pipeline (Fig 2) and all reward functions rely on these labels; label noise is acknowledged but only partially filtered.
  • domain assumption Spatial overlap (b-hat-prime approximately b-hat, s-hat-prime subset of s-hat) is a sufficient operationalization of evidence grounding.
    R_consistency rewards only coordinate-level matches; the paper equates verifiable evidence binding with box/span overlap, which may not capture semantic or causal grounding.
  • domain assumption The annotation MLLM (Qwen3-VL-32B), conditioned on ground-truth findings, produces physically consistent evidence reports that the filtering pipeline can clean.
    The 50K dataset is generated by a black-box model; correctness of the reasoning text is vetted by automated and peer-review filters but not by human forensic experts at scale.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evidence-Grounded Forensic Reasoning for Detecting and Grounding Multi-Modal Media Manipulation." pith.science (2026). https://pith.science/paper/XQJSXLCQ

@misc{pith2026260808009,
  author       = {Pith},
  title        = {Pith review of: Evidence-Grounded Forensic Reasoning for Detecting and Grounding Multi-Modal Media Manipulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XQJSXLCQ}},
  note         = {Machine review of arXiv:2608.08009}
}
read the original abstract

Fake news increasingly relies on cross-modal image-text forgeries, making transparent and verifiable reasoning chains an urgent need for Detecting and Grounding Multi-Modal Media Manipulation (DGM4). Existing methods produce black-box detection results without any decision rationale, limiting their reliability in forensic practice. Multi-modal Large Language Models (MLLMs) offer a natural path toward explainability, but applying them to DGM4 raises two difficulties. First, models tend to generate explanations disconnected from predicted evidence locations, producing unverified attribution. Second, enforcing evidence-conclusion consistency requires active optimization, yet uniform training signals fail to distinguish localization tokens from classification tokens, making multi-head joint training unreliable. We propose a multi-modal manipulation detector based on an Evidence-Grounded Forensic Reasoning (EFR) framework. EFR introduces an Anchor-and-Verify reasoning chain that enforces modality-isolated perception before cross-modal comparison, with conclusion coordinates as explicit anchors to which downstream evidence must spatially correspond. A verifiable reward system then enforces evidence-conclusion consistency during training, while a Modality-Decoupled Advantage (MDA) routing mechanism mitigats credit misassignment across prediction tasks. Experiments show that EFR achieves state-of-the-art performance while producing structured forensic reasoning records that explicitly bind explanations to evidence.

Figures

Figures reproduced from arXiv: 2608.08009 by the authors.

Figure 1
Figure 1. Comparison of existing approaches and our EFR [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The forensic reasoning annotation data construction pipeline. Raw DGM4 samples (208K) are filtered and stratified by [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overview of the two-stage EFR training framework. Stage 1 fine-tunes a base MLLM on 50K curated forensic reasoning [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Forensic reasoning output of EFR on a compound [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 24 canonical work pages

  1. [1]

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al . 2025. Qwen3-vl technical report.arXiv preprint arXiv:2511.21631(2025)

    arXiv 2025

  2. [2]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners.Advances in neural information processing systems33 (2020), 1877–1901

    2020

  3. [3]

    Daniele Cardullo, Simone Teglia, and Irene Amerini. 2025. LADLE-MM: Limited Annotation based Detector with Learned Ensembles for Multimodal Misinforma- tion.arXiv preprint arXiv:2512.20257(2025)

    work page arXiv 2025

  4. [4]

    Tao Chen, Jingyi Zhang, Decheng Liu, and Chunlei Peng. 2025. Mgffd-vlm: Multi- granularity prompt learning for face forgery detection with vlm.arXiv preprint arXiv:2507.12232(2025)

    arXiv 2025

  5. [5]

    Yueying Gao, Dongliang Chang, Bingyao Yu, Haotian Qin, Lei Chen, Kongming Liang, and Zhanyu Ma. 2025. Fakereasoning: Towards generalizable forgery detection and reasoning.arXiv e-prints(2025), arXiv–2503

    work page 2025

  6. [6]

    Olga Golovneva, Moya Peng Chen, Spencer Poff, Martin Corredor, Luke Zettle- moyer, Maryam Fazel-Zarandi, and Asli Celikyilmaz. 2023. ROSCOE: A Suite of Metrics for Scoring Step-by-Step Reasoning. InProceedings of the International Conference on Learning Representations (ICLR)

    work page 2023

  7. [7]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Peiyi Wang, Qihao Zhu, Runxin Xu, Ruoyu Zhang, Shirong Ma, Xiao Bi, et al. 2025. Deepseek-r1: Incen- tivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948(2025)

    arXiv 2025

  8. [8]

    Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. 2021. DeBERTa: Decoding-Enhanced BERT with Disentangled Attention. InProceedings of the International Conference on Learning Representations (ICLR)

    work page 2021

Show all 60 references
  1. [9]

    Xinan He, Yue Zhou, Bing Fan, Bin Li, Guopu Zhu, and Feng Ding. 2025. Vlforgery face triad: Detection, localization and attribution via multimodal large language models.arXiv preprint arXiv:2503.06142(2025)

    2025 arXiv

  2. [10]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.Iclr1, 2 (2022), 3

    2022

  3. [11]

    Zhenglin Huang, Jinwei Hu, Xiangtai Li, Yiwei He, Xingyu Zhao, Bei Peng, Baoyuan Wu, Xiaowei Huang, and Guangliang Cheng. 2025. Sida: Social media image deepfake detection, localization and explanation with large multimodal model. InProceedings of the Computer Vision and Patte...

    2025

  4. [12]

    Shan Jia, Reilin Lyu, Kangran Zhao, Yize Chen, Zhiyuan Yan, Yan Ju, Chuanbo Hu, Xin Li, Baoyuan Wu, and Siwei Lyu. 2024. Can chatgpt detect deepfakes? a study of using multimodal large language models for media forensics. InProceedings of the IEEE/CVF Conference on Computer Vi...

    2024

  5. [13]

    Jian-Yu Jiang-Lin, Kang-Yang Huang, et al. 2026. TriDF: Evaluating Perception, Detection, and Hallucination for Interpretable DeepFake Detection.Proceedings of the IEEE/CVF conference on computer vision and pattern recognition(2026)

    2026

  6. [14]

    Rohit Kundu, Shan Jia, Vishal Mohanty, Athula Balachandran, and Amit K Roy- Chowdhury. 2025. TruthLens: Visual Grounding for Universal DeepFake Reason- ing.arXiv preprint arXiv:2503.15867(2025)

    2025 arXiv

  7. [15]

    Jiwei Li, Michel Galley, Chris Brockett, Jianfeng Gao, and Bill Dolan. 2016. A Diversity-Promoting Objective Function for Neural Conversation Models. In Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics (NAACL)

    2016

  8. [16]

    Qilei Li, Mingliang Gao, Guisheng Zhang, Wenzhe Zhai, Jinyong Chen, and Gwanggil Jeon. 2024. Towards multimodal disinformation detection by vision- language knowledge interaction.Information Fusion102 (2024), 102037

    2024

  9. [17]

    Tianxiao Li, Zhenglin Huang, Haiquan Wen, Yiwei He, Shuchang Lyu, Baoyuan Wu, and Guangliang Cheng. 2025. RAIDX: A Retrieval-Augmented Generation and GRPO Reinforcement Learning Framework for Explainable Deepfake De- tection. InProceedings of the 33rd ACM International Confere...

    2025

  10. [18]

    Yiheng Li, Yang Yang, Zichang Tan, Huan Liu, Weihua Chen, Xu Zhou, and Zhen Lei. 2025. Unleashing the Potential of Consistency Learning for Detecting and Grounding Multi-Modal Media Manipulation. InProceedings of the Computer Vision and Pattern Recognition Conference. 9242–9252

    2025

  11. [19]

    Chin-Yew Lin. 2004. ROUGE: A Package for Automatic Evaluation of Summaries. InText Summarization Branches Out: Proceedings of the ACL Workshop

    2004

  12. [20]

    Kaiqing Lin, Zhiyuan Yan, Ruoxin Chen, Junyan Ye, Ke-Yue Zhang, Yue Zhou, Peng Jin, Bin Li, Taiping Yao, and Shouhong Ding. 2025. Seeing before reasoning: A unified framework for generalizable and explainable fake image detection. arXiv preprint arXiv:2509.25502(2025)

    2025

  13. [21]

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024. Improved baselines with visual instruction tuning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 26296–26306

    2024

  14. [22]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. Visual in- struction tuning.Advances in neural information processing systems36 (2023), 34892–34916

    2023

  15. [23]

    Huan Liu, Zichang Tan, Qiang Chen, Yunchao Wei, Yao Zhao, and Jingdong Wang. 2025. Unified frequency-assisted transformer framework for detecting and grounding multi-modal manipulation.International Journal of Computer Vision133, 3 (2025), 1392–1409

    2025

  16. [24]

    Jiawei Liu, Fanrui Zhang, Jiaying Zhu, Esther Sun, Qiang Zhang, and Zheng-Jun Zha. 2024. Forgerygpt: Multimodal large language model for explainable image forgery detection and localization.arXiv preprint arXiv:2410.10238(2024)

    2024 arXiv

  17. [25]

    Runxin Liu, Tian Xie, Jiaming Li, Lingyun Yu, and Hongtao Xie. 2025. IDseq: Decoupled and Sequentially Detecting and Grounding Multi-Modal Media Ma- nipulation. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 496–504

    2025

  18. [26]

    Shih-Yang Liu, Xin Dong, Ximing Lu, Shizhe Diao, Peter Belcak, Mingjie Liu, Min-Hung Chen, Hongxu Yin, Yu-Chiang Frank Wang, Kwang-Ting Cheng, et al

  19. [27]

    Qiong Nan, Juan Cao, Yongchun Zhu, Yanyan Wang, and Jintao Li. 2021. MD- FEND: Multi-domain fake news detection. InProceedings of the 30th ACM inter- national conference on information & knowledge management. 3343–3347

    2021

  20. [28]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback.Advances in neural information processing systems35 (...

    2022

  21. [29]

    Gan Pei, Jiangning Zhang, Menghan Hu, Zhenyu Zhang, Chengjie Wang, Yun- sheng Wu, Guangtao Zhai, Jian Yang, and Dacheng Tao. 2024. Deepfake genera- tion and detection: A benchmark and survey.Comput. Surveys(2024)

    2024

  22. [30]

    Lixiong Qin, Yang Zhang, Mei Wang, Jiani Hu, Weihong Deng, and Weiran Xu

  23. [31]

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. 2023. Direct preference optimization: Your language model is secretly a reward model.Advances in neural information processing systems36 (2023), 53728–53741

    2023

  24. [32]

    Andreas Rössler, Davide Cozzolino, Luisa Verdoliva, Christian Riess, Justus Thies, and Matthias Nießner. 2018. Faceforensics: A large-scale video dataset for forgery detection in human faces.arXiv preprint arXiv:1803.09179(2018)

    2018 arXiv

  25. [33]

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov

  26. [34]

    Rui Shao, Tianxing Wu, and Ziwei Liu. 2023. Detecting and grounding multi- modal media manipulation. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition. 6904–6913

    2023

  27. [35]

    Rui Shao, Tianxing Wu, Jianlong Wu, Liqiang Nie, and Ziwei Liu. 2024. Detecting and grounding multi-modal media manipulation and beyond.IEEE Transactions on Pattern Analysis and Machine Intelligence46, 8 (2024), 5556–5574

    2024

  28. [36]

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, et al. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models.arXiv preprint arXiv:2402.03300(2024)

    2024 arXiv

  29. [37]

    Jinjie Shen, Yaxiong Wang, Lechao Cheng, Nan Pu, and Zhun Zhong. 2025. Be- yond Artificial Misalignment: Detecting and Grounding Semantic-Coordinated Multimodal Manipulations. InProceedings of the 33rd ACM International Confer- ence on Multimedia. 11308–11317

    2025

  30. [38]

    Zhuokang Shen, Kaisen Zhang, Bohan Jia, Heming Jia, Yuan Fang, Zhou Yu, and Shaohui Lin. 2025. DF-LLaVA: Unlocking MLLMs for Synthetic Image Detection via Knowledge Injection and Conflict-Driven Self-Reflection.arXiv preprint arXiv:2509.14957(2025)

    2025

  31. [39]

    Gagandeep Singh, Samudi Amarsinghe, Priyanka Singh, and Xue Li. 2025. DGM4+: Dataset Extension for Global Scene Inconsistency.arXiv preprint arXiv:2509.26047(2025)

    2025

  32. [40]

    Hao Tan, Jun Lan, Zichang Tan, Ajian Liu, Chuanbiao Song, Senyuan Shi, Huijia Zhu, Weiqiang Wang, Jun Wan, and Zhen Lei. 2025. Veritas: Generalizable deepfake detection via pattern-aware reasoning.arXiv preprint arXiv:2508.21048 (2025)

    2025

  33. [41]

    Jiazhen Wang, Bin Liu, Changtao Miao, Zhiwei Zhao, Wanyi Zhuang, Qi Chu, and Nenghai Yu. 2024. Exploiting modality-specific features for multi-modal manipulation detection and grounding. InICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processin...

    2024

  34. [42]

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. 2024. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint MM ’26, November 10–14, 2026, Rio de Janeiro,...

    2024 arXiv

  35. [43]

    Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, et al. 2025. Internvl3. 5: Ad- vancing open-source multimodal models in versatility, reasoning, and efficiency. arXiv preprint arXiv:2508.18265(2025)

    2025 arXiv

  36. [44]

    Wenbo Xu, Wei Lu, Xiangyang Luo, and Jiantao Zhou. 2026. MARE: Multimodal Alignment and Reinforcement for Explainable Deepfake Detection via Vision- Language Models.arXiv preprint arXiv:2601.20433(2026)

    2026

  37. [45]

    Zhipei Xu, Xuanyu Zhang, Runyi Li, Zecheng Tang, Qing Huang, and Jian Zhang

  38. [46]

    Qiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan, Xiaochen Zuo, Yu Yue, Weinan Dai, Tiantian Fan, Gaohong Liu, Lingjun Liu, et al. 2025. Dapo: An open- source llm reinforcement learning system at scale.arXiv preprint arXiv:2503.14476 (2025)

    2025 arXiv

  39. [47]

    Xinquan Yu, Wei Lu, and Xiangyang Luo. 2025. Fine-grained Multiple Supervisory Network for Multi-modal Manipulation Detecting and Grounding.arXiv preprint arXiv:2508.02479(2025)

    2025 arXiv

  40. [48]

    Xinquan Yu, Wei Lu, Xiangyang Luo, and Rui Yang. 2026. CIEC: Coupling Implicit and Explicit Cues for Multimodal Weakly Supervised Manipulation Localization. arXiv preprint arXiv:2602.02175(2026)

    2026

  41. [49]

    Xiaojiang Zhang, Jinghui Wang, Zifei Cheng, Wenhao Zhuang, Zheng Lin, Min- glei Zhang, Shaojie Wang, Yinghan Cui, Chao Wang, Junyi Peng, et al. 2025. Srpo: A cross-domain implementation of large-scale reinforcement learning on llm. arXiv preprint arXiv:2504.14286(2025)

    2025 arXiv

  42. [50]

    Yuchen Zhang, Yaxiong Wang, Yujiao Wu, Lianwei Wu, Li Zhu, and Zhedong Zheng. 2025. The Coherence Trap: When MLLM-Crafted Narratives Exploit Manipulated Visual Contexts.arXiv preprint arXiv:2505.17476(2025)

    2025

  43. [51]

    Zhenxing Zhang, Yaxiong Wang, Lechao Cheng, Zhun Zhong, Dan Guo, and Meng Wang. 2025. Asap: Advancing semantic alignment promotes multi-modal manipulation detecting and grounding. InProceedings of the Computer Vision and Pattern Recognition Conference. 4005–4014

    2025

  44. [52]

    Hanqing Zhao, Wenbo Zhou, Dongdong Chen, Tianyi Wei, Weiming Zhang, and Nenghai Yu. 2021. Multi-attentional deepfake detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2185–2194

    2021

  45. [53]

    Chujie Zheng, Shixuan Liu, Mingze Li, Xiong-Hui Chen, Bowen Yu, Chang Gao, Kai Dang, Yuqiong Liu, Rui Men, An Yang, et al. 2025. Group sequence policy optimization.arXiv preprint arXiv:2507.18071(2025)

    2025 arXiv

  46. [54]

    Ziyin Zhou, Yunpeng Luo, Yuanchen Wu, Ke Sun, Jiayi Ji, Ke Yan, Shouhong Ding, Xiaoshuai Sun, Yunsheng Wu, and Rongrong Ji. 2025. Aigi-holmes: Towards explainable and generalizable ai-generated image detection via multimodal large language models. InProceedings of the IEEE/CVF...

    2025

  47. [55]

    Tinghui Zhu, Qin Liu, Fei Wang, Zhengzhong Tu, and Muhao Chen. 2024. Unrav- eling cross-modality knowledge conflicts in large vision-language models.arXiv preprint arXiv:2410.03659(2024)

    2024 arXiv

  48. [56]

    Zhihao Zhu, Jiafeng Liang, Shixin Jiang, Jinlan Fu, Ming Liu, Guanglu Sun, See- Kiong Ng, and Bing Qin. 2026. Analyzing Reasoning Consistency in Large Multimodal Models under Cross-Modal Conflicts.arXiv preprint arXiv:2601.04073 (2026)

    2026

  49. [2017]

    Proximal policy optimization algorithms.arXiv preprint arXiv:1707.06347 (2017)

    2017 arXiv

  50. [2024]

    Fakeshield: Explainable image forgery detection and localization via multi- modal large language models.arXiv preprint arXiv:2410.02761(2024)

    2024 arXiv

  51. [2025]

    arXiv preprint arXiv:2510.20531(2025)

    Fake-in-Facext: Towards Fine-Grained Explainable DeepFake Analysis. arXiv preprint arXiv:2510.20531(2025)

    2025

  52. [2026]

    Gdpo: Group reward-decoupled normalization policy optimization for multi-reward rl optimization.arXiv preprint arXiv:2601.05242(2026)

    2026 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.