Summary
Short videos on finance platforms sometimes trick people into scams using quick visual signals and text that are hard to catch. The authors found that relying only on text detection misses many of these clues because not all warnings are written out. They combined text reading with an AI that understands both words and images together, which better spots and pinpoints scam content in these videos. They also created a dataset to train and test their method and designed a new scoring system to handle tricky class imbalances. Their approach finds more fraud signals and locates their timing more precisely compared to text-only methods.
What this means in practice
- •For content moderation teams: Detect and localize fraudulent inducements in short finance videos more accurately by combining text and visual cues with multimodal AI models.
- •For social media platform engineers: Enhance video moderation pipelines by integrating multimodal models that improve recall and timing of suspicious content beyond text-only systems.
Abstract
Short video platforms have become a primary channel for black and gray industry operations in internet finance, where inducements such as prompts to like, follow, join private chat groups, or click external links frequently steer users toward fraud. Because such videos are short, dynamic, and often convey intent through fleeting visual cues rather than stable text, a moderation system must guarantee detection recall and localize the offending segment so reviewers inspect only the relevant clip. Existing OCR+LLM pipelines transcribe on-screen text frame by frame and classify the transcript, but exploit only textual information. Many inducements are conveyed through visual cues OCR cannot observe, and keyframe-based localization cannot recover precise temporal boundaries. We explore OCR+MLLM, which augments the text pipeline with a Multimodal Large Language Model, and compare three routes: OCR+LLM, OCR+MLLM, and a large zero-shot MLLM, along detection, latency, token consumption, and deployment complexity. OCR preserves the text signal that guarantees recall, while the MLLM jointly performs OCR-grounded discrimination and temporal grounding, localizing both textual and non-textual inducements. We construct a dataset of 2,419 internet finance short videos (66.49 hours) with 3,194 annotated events across four categories, and study reward design for RL-based grounding under class imbalance. We report two findings: auxiliary reward terms are exploited through reward hacking, and stronger localization rewards degrade class-balanced detection. We therefore design a recall-weighted F2 localization reward with class-balanced resampling. Experiments show OCR+MLLM localizes better than the keyframe OCR+LLM baseline (mIoU 26.52 vs. 14.49) and detects non-textual inducements that text-only pipelines miss, while maintaining 67.0 percent macro recall.