Beyond Context: Large Language Models' Failure to Grasp Users' Intent

2025/12/24 by Ahmed M. Hussain, Hussain, Ahmed M., Salahuddin Salahuddin +2 · 1 voice
Computer Science · Medicine · Social Sciences · #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #Artificial Intelligence in Healthcare and Education #Computation and Language (cs.CL) #Computers and Society (cs.CY) #Cryptography and Security (cs.CR) #Ethics and Social Impacts of AI #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2512.21110

openalex publication_date 2025/12/24 · openalex created_date 2025/12/26 · openalex updated_date 2026/07/28

Abstract

Current Large Language Models (LLMs) safety approaches focus on explicitly harmful content while overlooking a critical vulnerability: the inability to understand context and recognize user intent. This creates exploitable vulnerabilities that malicious users can systematically leverage to circumvent safety mechanisms. We empirically evaluate multiple state-of-the-art LLMs, including ChatGPT, Claude, Gemini, and DeepSeek. Our analysis demonstrates the circumvention of reliable safety mechanisms through emotional framing, progressive revelation, and academic justification techniques. Notably, reasoning-enabled configurations amplified rather than mitigated the effectiveness of exploitation, increasing factual precision while failing to interrogate the underlying intent. The exception was Claude Opus 4.1, which prioritized intent detection over information provision in some use cases. This pattern reveals that current architectural designs create systematic vulnerabilities. These limitations require paradigmatic shifts toward contextual understanding and intent recognition as core safety capabilities rather than post-hoc protective mechanisms.

Citations

• Jailbreaking Large Language Models Through Content Concretization
• Monitoring Decomposition Attacks in LLMs with Lightweight Sequential Monitors
• Red Teaming the Mind of the Machine: A Systematic Evaluation of Prompt Injection and Jailbreak Vulnerabilities in LLMs
• Prompt, Divide, and Conquer: Bypassing Large Language Model Safety Filters via Segmented and Distributed Prompt Processing
• Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
• LongSafety: Evaluating Long-Context Safety of Large Language Models
• Is Safety Standard Same for Everyone? User-Specific Safety Evaluation of Large Language Models
• CASE-Bench: Context-Aware SafEty Benchmark for Large Language Models
• CySecBench: Generative AI-based CyberSecurity-focused Prompt Dataset for Benchmarking Large Language Models
• Intent Detection in the Age of LLMs
• Evaluating Very Long-Term Conversational Memory of LLM Agents
• DrAttack: Prompt Decomposition and Reconstruction Makes Powerful LLM Jailbreakers
• HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal
• Intention Analysis Makes LLMs A Good Jailbreak Defender
• Attack Prompt Generation for Red Teaming and Defending Large Language Models
• AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models
• SafetyBench: Evaluating the Safety of Large Language Models
• "Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models
• Code of "Sirens' Whisper: Inaudible Near-Ultrasonic Jailbreaks of Speech-Driven LLMs"
• Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
• Lost in the Middle: How Language Models Use Long Contexts
• Jailbroken: How Does LLM Safety Training Fail?
• DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models
• Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models
• Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study
• GPT-4 Technical Report
• Not What You've Signed Up For: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection
• Affective Coherence Monitoring for Transformer-Based Language Models
• Ignore Previous Prompt: Attack Techniques For Language Models
• Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned
• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Training Compute-Optimal Large Language Models
• Context is Everything: Implicit Identification for Dynamics Adaptation
• Training language models to follow instructions with human feedback
• Red Teaming Language Models with Language Models
• Ethical and social risks of harm from Language Models
• Unsolved Problems in ML Safety
• On the Opportunities and Risks of Foundation Models
• The Values Encoded in Machine Learning Research
• Extracting Training Data from Large Language Models
• RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
• Language Models are Few-Shot Learners
• Recipes for building an open-domain chatbot
• Longformer: The Long-Document Transformer
• Efficient Intent Detection with Dual Sentence Encoders
• A Primer in BERTology: What we know about how BERT works
• The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence
• A Simple Framework for Contrastive Learning of Visual Representations
• Scaling Laws for Neural Language Models
• Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing
• Learning by Cheating
• Compressive Transformers for Long-Range Sequence Modelling
• Adversarial NLI: A New Benchmark for Natural Language Understanding
• Fine-Tuning Language Models from Human Preferences
• Towards Scalable Multi-domain Conversational Agents: The Schema-Guided\n Dialogue Dataset
• Universal Adversarial Triggers for Attacking and Analyzing NLP
• VisualBERT: A Simple and Performant Baseline for Vision and Language
• ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks
• Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment
• A Multiscale Visualization of Attention in the Transformer Model
• What Does BERT Look At? An Analysis of BERT's Attention
• Do Neural Dialog Systems Use the Conversation History Effectively? An Empirical Study
• Pretraining Methods for Dialog Context Representation Learning
• Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference
• Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
• Deep contextualized word representations
• Graph Attention Networks
• Adversarial Examples for Evaluating Reading Comprehension Systems
• Deep reinforcement learning from human preferences
• Attention Is All You Need
• Semi-Supervised Classification with Graph Convolutional Networks
• Concrete Problems in AI Safety
• Explaining and Harnessing Adversarial Examples
• Intriguing properties of neural networks
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Discussions

• Beyond Context: Large Language Models Failure to Grasp Users Intent [hn, 4 points, 0 comments]

Related

• Exploring the Vulnerability of the Content Moderation Guardrail in Large Language Models via Intent Manipulation
• Unveiling the Misuse Potential of Base Large Language Models via In-Context Learning
• IntentMiner: Intent Inversion Attack via Tool Call Analysis in the Model Context Protocol
• Large Language Models are Vulnerable to Bait-and-Switch Attacks for Generating Harmful Content
• Beyond the Safeguards: Exploring the Security Risks of ChatGPT