Measuring and Evaluating the Performance of Generative AI Models for Scam Detection

2026-07-19

AI Summary

This study investigates the capability of general-purpose large language models (LLMs) to detect diverse real-world online scams without fine-tuning. To this end, we construct the first benchmark dataset comprising authentic scam instances spanning multiple formats and themes, and systematically evaluate nine LLMs of varying scales and architectures under diverse prompting strategies, comparing their performance against fine-tuned BERT classifiers. Our results demonstrate that well-designed prompt engineering substantially enhances the performance of smaller models. Overall, LLMs outperform fine-tuned baselines, particularly exhibiting superior generalization on previously unseen scam types, thereby validating the effectiveness of their pre-trained knowledge for security-related tasks.

Abstract

Online scams continue to cause substantial financial and personal harm. As a result, detection systems based on Large Language Models (LLMs) have been integrated into security products ranging from email gateways and browser extensions to fraud-monitoring dashboards. As this adoption accelerates, a common belief has taken hold: that these models are broadly suitable for scam detection. In this work, we investigate whether LLMs, with their strong capabilities in understanding intent, context, and reasoning, can effectively detect scams across diverse scenarios without task-specific fine-tuning. We curate and release a unique benchmark dataset of real-world scams spanning multiple formats and topics. We evaluate nine LLMs of varying sizes and architectures, examining their performance under different prompting strategies and comparing them to a fine-tuned BERT-based classifier. Our results show that while larger LLMs generally outperform smaller ones, effective prompting substantially boosts the performance of smaller models. Moreover, LLMs are better at generalizing to unseen scams compared to fine-tuned models, suggesting that pre-trained knowledge contributes meaningfully to scam detection. We release our dataset and evaluation framework to facilitate future research in robust scam detection using language models.

Problem

• scam detection
• Large Language Models
• generative AI
• benchmark dataset
• model evaluation

Innovation

• generative AI
• scam detection
• large language models
• prompting strategies
• benchmark dataset