Analysis of Threat-Based Manipulation in Large Language Models: A Dual Perspective on Vulnerabilities and Performance Enhancement Opportunities

TL;DR

This paper analyzes how large language models respond to threat-based manipulations, revealing vulnerabilities and opportunities for performance improvements, with implications for AI safety and prompt engineering.

Contribution

It introduces a new threat taxonomy and multi-metric evaluation framework to systematically assess vulnerabilities and performance gains in LLMs.

Findings

Systematic vulnerabilities identified across models and tasks.

Significant performance improvements with effect sizes up to +1336%.

Policy evaluation under role-based threats shows high metric significance.

Abstract

Large Language Models (LLMs) demonstrate complex responses to threat-based manipulations, revealing both vulnerabilities and unexpected performance enhancement opportunities. This study presents a comprehensive analysis of 3,390 experimental responses from three major LLMs (Claude, GPT-4, Gemini) across 10 task domains under 6 threat conditions. We introduce a novel threat taxonomy and multi-metric evaluation framework to quantify both negative manipulation effects and positive performance improvements. Results reveal systematic vulnerabilities, with policy evaluation showing the highest metric significance rates under role-based threats, alongside substantial performance enhancements in numerous cases with effect sizes up to +1336%. Statistical analysis indicates systematic certainty manipulation (pFDR < 0.0001) and significant improvements in analytical depth and response quality. These findings have dual implications for AI safety and practical prompt engineering in high-stakes applications.