MALicious INTent Dataset and Inoculating LLMs for Enhanced Disinformation Detection

arXiv cs.CL / 3/17/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

MALINT is the first human-annotated English corpus capturing disinformation and malicious intent, developed with expert fact-checkers.
The work benchmarks 12 language models, including small models like BERT and large models such as Llama 3.3, on binary and multilabel intent classification tasks.
It proposes intent-based inoculation, an intent-augmented reasoning approach for LLMs to mitigate the persuasive impact of disinformation by integrating intent analysis.
The authors demonstrate that intent-augmented reasoning improves zero-shot disinformation detection across six datasets, five LLMs, and seven languages, and they release the MALINT dataset with annotations.

Abstract

The intentional creation and spread of disinformation poses a significant threat to public discourse. However, existing English datasets and research rarely address the intentionality behind the disinformation. This work presents MALINT, the first human-annotated English corpus developed in collaboration with expert fact-checkers to capture disinformation and its malicious intent. We utilize our novel corpus to benchmark 12 language models, including small language models (SLMs) such as BERT and large language models (LLMs) like Llama 3.3, on binary and multilabel intent classification tasks. Moreover, inspired by inoculation theory from psychology and communication studies, we investigate whether incorporating knowledge of malicious intent can improve disinformation detection. To this end, we propose intent-based inoculation, an intent-augmented reasoning for LLMs that integrates intent analysis to mitigate the persuasive impact of disinformation. Analysis on six disinformation datasets, five LLMs, and seven languages shows that intent-augmented reasoning improves zero-shot disinformation detection. To support research in intent-aware disinformation detection, we release the MALINT dataset with annotations from each annotation step.