AttackEval: A Systematic Empirical Study of Prompt Injection Attack Effectiveness Against Large Language Models

TL;DR

This paper systematically evaluates prompt injection attack strategies against large language models, revealing vulnerabilities and guiding more robust defense designs.

Contribution

It introduces AttackEval, a comprehensive taxonomy and empirical framework for analyzing prompt injection attack effectiveness and defense resilience.

Findings

Obfuscation achieves the highest attack success rate (ASR=0.76) against defenses.

Semantic/Social attacks maintain high ASR (0.44-0.48) against intent-aware defenses.

Combining attack strategies dramatically increases ASR, e.g., OBF + EM reaches 97.6%.

Abstract

Prompt injection has emerged as a critical vulnerability in large language model (LLM) deployments, yet existing research is heavily weighted toward defenses. The attack side -- specifically, which injection strategies are most effective and why -- remains insufficiently studied.We address this gap with AttackEval, a systematic empirical study of prompt injection attack effectiveness. We construct a taxonomy of ten attack categories organized into three parent groups (Syntactic, Contextual, and Semantic/Social), populate each category with 25 carefully crafted prompts (250 total), and evaluate them against a simulated production victim system under four progressively stronger defense tiers. Experiments reveal several non-obvious findings: (1) Obfuscation (OBF) achieves the highest single-attack success rate (ASR = 0.76) against even intent-aware defenses, because it defeats both keyword matching and semantic similarity checks simultaneously; (2) Semantic/Social attacks - Emotional Manipulation (EM) and Reward Framing (RF) - maintain high ASR (0.44-0.48) against intent-aware defenses due to their natural language surface, which evades structural anomaly detection; (3) Composite attacks combining two complementary strategies dramatically boost ASR, with the OBF + EM pair reaching 97.6%; (4) Stealth correlates positively with residual ASR against semantic defenses (r = 0.71), implying that future defenses must jointly optimize for both structural and behavioral signals. Our findings identify concrete blind spots in current defenses and provide actionable guidance for designing more robust LLM safety systems.