SCRIPTMIND: Crime Script Inference and Cognitive Evaluation for LLM-based Social Engineering Scam Detection System
This work addresses social engineering scam detection for users, offering a novel integration of automated reasoning and human cognition, though it is incremental in applying existing LLM methods to a specific domain.
The authors tackled the problem of detecting personalized social engineering scams by proposing ScriptMind, a framework that fine-tunes small LLMs for scam reasoning, which outperformed GPT-4o by 13% in detection accuracy and improved users' suspicion levels in simulations.
Social engineering scams increasingly employ personalized, multi-turn deception, exposing the limits of traditional detection methods. While Large Language Models (LLMs) show promise in identifying deception, their cognitive assistance potential remains underexplored. We propose ScriptMind, an integrated framework for LLM-based scam detection that bridges automated reasoning and human cognition. It comprises three components: the Crime Script Inference Task (CSIT) for scam reasoning, the Crime Script-Aware Inference Dataset (CSID) for fine-tuning small LLMs, and the Cognitive Simulation-based Evaluation of Social Engineering Defense (CSED) for assessing real-time cognitive impact. Using 571 Korean phone scam cases, we built 22,712 structured scammer-sequence training instances. Experimental results show that the 11B small LLM fine-tuned with ScriptMind outperformed GPT-4o by 13%, achieving superior performance over commercial models in detection accuracy, false-positive reduction, scammer utterance prediction, and rationale quality. Moreover, in phone scam simulation experiments, it significantly enhanced and sustained users' suspicion levels, improving their cognitive awareness of scams. ScriptMind represents a step toward human-centered, cognitively adaptive LLMs for scam defense.