From Cheap Fakes to Pure Synthesis: Addressing the New Era of T2V Fake News Videos

2026-08-06

AI Summary

This study addresses the challenge of detecting purely synthetic fake news videos generated by text-to-video (T2V) models, a task in which existing methods are vulnerable to single-modality shortcuts and semantic-visual inconsistencies. To tackle this issue, the work formalizes the detection task as a ternary classification problem—distinguishing real, cheapfake, and purely synthetic videos—and introduces PS-FNVD, the first dataset dedicated to purely synthetic fake news videos. The authors propose R-T2V, a reasoning-guided detection framework that integrates high-level semantic logic with low-level generation artifacts for authenticity assessment. By combining conditional rationale generation with supervised fine-tuning, R-T2V outperforms ten state-of-the-art baselines, achieving the highest accuracy and macro F1 score, surpassing the second-best method by 12.20 and 8.46 percentage points, respectively.

Abstract

Recent text-to-video (T2V) generation models enable fake news videos to be synthesized from scratch, shifting the threat beyond cheap fakes assembled from existing footage. Such news videos can closely match fabricated narratives, creating a modality alignment trap for existing detectors. Existing datasets lack pure synthesis fake news videos. Although directly prompting T2V models with descriptions of fake news videos can yield perfectly aligned samples, it reduces the fake news video detection (FNVD) to unimodal shortcuts and causes semantic-visual degeneration. To counter this, we formulate T2V-FNVD as a novel ternary classification task with three labels (real, cheap fake, and pure synthesis fake) and construct the first pure synthesis fake news video dataset (PS-FNVD). PS-FNVD includes fabricated events with aligned deception (Type 1) and true events with false visual provenance (Type 2), preventing models from exploiting unimodal shortcuts. Furthermore, we propose the Reasoning-guided T2V-FNVD (R-T2V) framework. Trained through conditioned rationale generation and supervised fine-tuning, R-T2V integrates high-level semantic logic with low-level physical generative traces to predict the ternary veracity label. Extensive experiments across 10 prevailing baselines show that R-T2V achieves the state-of-the-art performance, outperforming the second-best baseline by 12.20 percentage points in accuracy and 8.46 percentage points in macro $F_1$.

Problem

• fake news video detection
• text-to-video generation
• pure synthesis
• multimodal deception
• deepfake

Innovation

• text-to-video
• fake news detection
• pure synthesis
• ternary classification
• reasoning-guided framework