Multimodal fake news detection remains challenging because deceptive posts exhibit heterogeneous manipulation patterns, while most existing methods still rely on a unified fusion strategy. This mismatch limits their ability to adapt to different evidence preferences across samples, encourages entanglement between deception cues and topical semantics, and weakens decision making when textual, visual, and cross-modal signals conflict. To address these issues, we propose TERN, a type-aware evidence reasoning network for multimodal fake news detection. TERN induces latent deception types from image-side multimodal features through prototype-based clustering, uses the induced assignments as structural priors for downstream veracity prediction, disentangles type-discriminative factors from semantic content, and performs type-conditioned hierarchical reasoning over text semantics, image authenticity, and cross-modal consistency. Experiments on MR2-Chinese, MR2-English, Weibo, and PHEME show that TERN achieves an average accuracy of 93.21% and an average F1 score of 91.09% while also improving Matthews correlation coefficient over representative multimodal baselines.

Multimodal misinformation often reuses authentic images while shifting key contextual details such as time, location, identity, or event description, making verification harder than ordinary image-text matching. We study this problem under a bounded-evidence setting, where verification relies on limited external evidence and, when available, early social traces collected within a fixed post-publication window. We propose SA-GGCoT, a three-stage framework that uses checkpoint-aware state tracking to localize conflicting evidence, LLMs to convert selected evidence into structured rationales, and a heterogeneous evidence graph to verify evidence consistency. Experiments on Weibo, PHEME, and MR\(^2\) show consistent gains over protocol-matched baselines, with ablation and grounding analyses supporting the contribution of each component.

Multimodal fake news is often locally plausible but relationally fractured. The image may be authentic, the text fluent, and the named entities recognizable, yet the evidence chain connecting entity, context, visual content, and claim can be broken. Most multimodal detectors encode the observed post as a static image-text state. MEEL (Manipulation-Equivariant Evidence Learning) instead learns typed directions of manipulation-induced evidence-state movement under a fixed-evidence setting. The framework constructs controlled edited views through image swapping, entity shifting, context dropping, relation misbinding, and text overclaiming. Operation-specific displacement prototypes make each manipulation correspond to a predictable direction in evidence-state space. At inference time, no edited view is generated; veracity is predicted from the natural evidence state, its alignment with the learned manipulation directions, sparse direction profiles, and equivariance uncertainty. On MR2-Chinese, MR2-English, and Weibo, MEEL achieves F1/MCC scores of 96.25%/0.9275, 95.21%/0.9231, and 92.25%/0.8296, respectively. Stability analysis, ablations, and direction diagnostics identify semantic edit construction and equivariant direction learning as the main sources of the displacement signal; scalar cues, pseudo-news supervision, random-edit controls, and uncertainty modeling have auxiliary or diagnostic roles. The model learns the movement of evidence states under semantic manipulation, beyond static image-text compatibility.

Multimodal fake news is difficult to detect when text and image remain topically aligned but diverge in local factual details such as entities, locations, time, or event attributes. Existing methods often rely on global cross-modal fusion and may therefore overlook these localized contradictions. We present LCFND, a staged and fully specified bounded-evidence verification pipeline for localized conflict discovery, structured evidence generation, and graph-based verification. Optimal transport identifies high-conflict token-region pairs and constructs localized evidence packets. A parameter-efficient LLM trained with OT-grounded structured targets and training-split label-informed weak relation calibration then converts these packets into structured evidence, including claims, visual evidence, contradiction types, and grounded rationales. A heterogeneous evidence graph integrates OT conflict priors and LLM-generated evidence for veracity prediction, and all graph-construction rules, training objectives, and evaluation protocols are stated explicitly for reproducibility. Under a controlled matched-backbone protocol on Weibo, Twitter, and GossipCop, LCFND improves F1 over the top matched-backbone baseline MGCA by 1.47, 1.73, and 1.67 points, respectively. The gain remains positive when the visual backbone is strengthened to ViT-B/16 or Swin-T. Manual evidence evaluation on 500 samples per dataset reports 85.9–87.3% claim grounding and 78.1–80.2% contradiction-type correctness. Under the P3b matched-input protocol, LCFND remains above FND-LLM-M and MMRGV-lite, and improves average cross-dataset transfer F1 from 67.10 to 69.98. Full online inference takes 1.47 s per sample in our setting, so the framework is better suited to offline or high-risk verification than to large-scale real-time screening.

Deceptive multimodal posts often remain globally plausible because the text and image can appear mutually consistent while the veracity claim depends on a small set of fragile evidence supports. Existing multimodal fake news detectors have advanced fusion, fine-grained interaction, knowledge-enhanced reasoning, and counterfactual learning, but they mainly strengthen the representation of the observed post and rarely measure which support actually sustains the final decision. We propose SRFP, a typed structural-response framework that represents each post as a configuration of entity, background, relation, and local-visual support representations. For each support type, SRFP applies a controlled cue-specific perturbation, compares the resulting factual and support-conditioned alternative states, encodes the state shift as a response trajectory, and compresses multi-type responses into a collapse signature for prediction and mechanism analysis. SRFP obtains F1/MCC scores of 94.15%/88.69% on MR2_Chinese, 93.33%/88.99% on MR2_English, and 92.68%/83.71% on Weibo. Ablation and matched-control experiments support the importance of preserving the semantic correspondence among support type, support representation, perturbation operator, and response descriptor, while progressive support-quality analysis characterizes the robustness of the learned response. These results suggest that multimodal fake news detection benefits from observing structural response under typed support challenges in addition to building stronger static image-text representations.