{"total":81,"rows":[{"id":"2404.13208","title":"The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions","created":"2024-04-19","cited_by_count":493,"area":"safety"},{"id":"2406.14598","title":"SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal","created":"2024-06-20","cited_by_count":240,"area":"alignment"},{"id":"2410.05451","title":"SecAlign: Defending Against Prompt Injection with Preference Optimization","created":"2024-10-07","cited_by_count":210,"area":"alignment"},{"id":"2406.13352","title":"AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents","created":"2024-06-19","cited_by_count":145,"area":"agents"},{"id":"2411.00348","title":"Attention Tracker: Detecting Prompt Injection Attacks in LLMs","created":"2024-11-01","cited_by_count":95,"area":"safety"},{"id":"2405.05175","title":"AirGapAgent: Protecting Privacy-Conscious Conversational Agents","created":"2024-05-08","cited_by_count":94,"area":"agents"},{"id":"2602.20156","title":"Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks","created":"2026-02-23","cited_by_count":90,"area":"agents"},{"id":"2506.08837","title":"Design Patterns for Securing LLM Agents against Prompt Injections","created":"2025-06-10","cited_by_count":73,"area":"agents"},{"id":"2406.00799","title":"Get my drift? Catching LLM Task Drift with Activation Deltas","created":"2024-06-02","cited_by_count":72,"area":"rag"},{"id":"2410.09102","title":"Instructional Segment Embedding: Improving LLM Safety with Instruction Hierarchy","created":"2024-10-09","cited_by_count":67,"area":"safety"},{"id":"2503.24370","title":"Effectively Controlling Reasoning Models through Thinking Intervention","created":"2025-03-31","cited_by_count":65,"area":"lr"},{"id":"2502.08745","title":"IHEval: Evaluating Language Models on Following the Instruction Hierarchy","created":"2025-02-12","cited_by_count":53,"area":null},{"id":"2502.16580","title":"Can Indirect Prompt Injection Attacks Be Detected and Removed?","created":"2025-02-23","cited_by_count":52,"area":"safety"},{"id":"2505.21936","title":"RedTeamCUA: Realistic Adversarial Testing of Computer-Use Agents in Hybrid Web-OS Environments","created":"2025-05-28","cited_by_count":46,"area":"agents"},{"id":"2510.19207","title":"Defending Against Prompt Injection with DataFilter","created":"2025-10-22","cited_by_count":35,"area":"safety"},{"id":"2502.15851","title":"Control Illusion: The Failure of Instruction Hierarchies in Large Language Models","created":"2025-02-21","cited_by_count":33,"area":null},{"id":"2507.07974","title":"Defending Against Prompt Injection With a Few DefensiveTokens","created":"2025-07-10","cited_by_count":33,"area":"safety"},{"id":"2509.22830","title":"ChatInject: Abusing Chat Templates for Prompt Injection in LLM Agents","created":"2025-09-26","cited_by_count":31,"area":"agents"},{"id":"2510.12252","title":"PromptLocate: Localizing Prompt Injection Attacks","created":"2025-10-14","cited_by_count":28,"area":"safety"},{"id":"2506.01055","title":"Simple Prompt Injection Attacks Can Leak Personal Data Observed by LLM Agents During Task Execution","created":"2025-06-01","cited_by_count":26,"area":"agents"},{"id":"2503.10566","title":"ASIDE: Architectural Separation of Instructions and Data in Language Models","created":"2025-03-13","cited_by_count":24,"area":"safety"},{"id":"2410.21492","title":"FATH: Authentication-based Test-time Defense against Indirect Prompt Injection Attacks","created":"2024-10-28","cited_by_count":22,"area":"safety"},{"id":"2506.09956","title":"LLMail-Inject: A Dataset from a Realistic Adaptive Prompt Injection Challenge","created":"2025-06-11","cited_by_count":21,"area":"safety"},{"id":"2603.12277","title":"Prompt Injection as Role Confusion","created":"2026-02-22","cited_by_count":17,"area":"safety"},{"id":"2507.07417","title":"May I have your Attention? Breaking Fine-Tuning based Prompt Injection Defenses using Architecture-Aware Attacks","created":"2025-07-10","cited_by_count":15,"area":"safety"},{"id":"2512.01295","title":"Systems Security Foundations for Agentic Computing","created":"2025-12-01","cited_by_count":13,"area":"agents"},{"id":"2510.08829","title":"CommandSans: Securing AI Agents with Surgical Precision Prompt Sanitization","created":"2025-10-09","cited_by_count":12,"area":"agents"},{"id":"2604.08499","title":"PIArena: A Platform for Prompt Injection Evaluation","created":"2026-04-09","cited_by_count":11,"area":"safety"},{"id":"2507.02735","title":"Meta SecAlign: A Secure Foundation LLM Against Prompt Injection Attacks","created":"2025-07-03","cited_by_count":10,"area":"agents"},{"id":"2505.18907","title":"Stronger Enforcement of Instruction Hierarchy via Augmented Intermediate Representations","created":"2025-05-25","cited_by_count":9,"area":"safety"},{"id":"2512.00966","title":"Mitigating Indirect Prompt Injection via Instruction-Following Intent Analysis","created":"2025-11-30","cited_by_count":9,"area":"agents"},{"id":"2408.07004","title":"Casper: Prompt Sanitization for Protecting User Privacy in Web-Based Large Language Models","created":"2024-08-13","cited_by_count":8,"area":null},{"id":"2603.12230","title":"Security Considerations for Artificial Intelligence Agents","created":"2026-03-12","cited_by_count":8,"area":null},{"id":"2606.10525","title":"Assessing Automated Prompt Injection Attacks in Agentic Environments","created":"2026-06-09","cited_by_count":7,"area":"agents"},{"id":"2505.00626","title":"The Illusion of Role Separation: Hidden Shortcuts in LLM Role Learning (and How to Fix Them)","created":"2025-05-01","cited_by_count":6,"area":null},{"id":"2602.13597","title":"AlignSentinel: Alignment-Aware Detection of Prompt Injection Attacks","created":"2026-02-14","cited_by_count":5,"area":"safety"},{"id":"2605.18991","title":"Agent Security is a Systems Problem","created":"2026-05-18","cited_by_count":5,"area":null},{"id":"2505.14549","title":"Can Large Language Models Really Recognize Your Name?","created":"2025-05-20","cited_by_count":4,"area":null},{"id":"2601.11199","title":"SD-RAG: A Prompt-Injection-Resilient Framework for Selective Disclosure in Retrieval-Augmented Generation","created":"2026-01-16","cited_by_count":4,"area":"rag"},{"id":"2407.03391","title":"Soft Begging: Modular and Efficient Shielding of LLMs against Prompt Injection and Jailbreaking based on Prompt Tuning","created":"2024-07-03","cited_by_count":3,"area":"safety"},{"id":"2504.16871","title":"Exploring How LLMs Capture and Represent Domain-Specific Knowledge","created":"2025-04-23","cited_by_count":3,"area":null},{"id":"2605.12460","title":"Multi-Stream LLMs: Unblocking Language Models with Parallel Streams of Thoughts, Inputs and Outputs","created":"2026-05-12","cited_by_count":3,"area":"lr"},{"id":"2606.26479","title":"Adaptive Evaluation of Out-of-Band Defenses Against Prompt Injection in LLM Agents","created":"2026-06-25","cited_by_count":3,"area":"agents"},{"id":"2608.21500","title":"SecOPD: Mitigating Adaptive Prompt Injections by On-Policy Distillation","created":"2026-08-21","cited_by_count":3,"area":"safety"},{"id":"2505.15738","title":"Checkpoint-GCG: Auditing and Attacking Fine-Tuning-Based Prompt Injection Defenses","created":"2025-05-21","cited_by_count":2,"area":"agents"},{"id":"2507.02699","title":"Your Mailbox Is Mine: Prompt Injection Attacks Against Real-World LLM Email Agents","created":"2025-07-03","cited_by_count":2,"area":"agents"},{"id":"2602.01795","title":"RedVisor: Reasoning-Aware Prompt Injection Defense via Zero-Copy KV Cache Reuse","created":"2026-02-02","cited_by_count":2,"area":"lr"},{"id":"2605.12233","title":"No More, No Less: Task Alignment in Terminal Agents","created":"2026-05-12","cited_by_count":2,"area":null},{"id":"2606.30783","title":"Security--Fidelity Tradeoffs: The Hidden Cost of Prompt Injection Defense","created":"2026-06-29","cited_by_count":2,"area":"safety"},{"id":"2608.17597","title":"HarnessRisk: A Lifecycle-Oriented Benchmark for Agent Harness Safety","created":"2026-08-18","cited_by_count":2,"area":"agents"},{"id":"2602.07104","title":"Extended to Reality: Prompt Injection in 3D Environments","created":"2026-02-06","cited_by_count":1,"area":"mllm"},{"id":"2606.04109","title":"Discourse-Role Labels as Presentation-Time Variables for Context Use in Language Models","created":"2026-06-02","cited_by_count":1,"area":null},{"id":"2606.09005","title":"Document-Authored Control-Signal Impersonation: A Low-Cost Indirect Prompt Attack on RAG Safety Boundaries","created":"2026-06-08","cited_by_count":1,"area":"rag"},{"id":"2606.27567","title":"On the Inseparability of Instructions and Data in Shared-Embedding Sequence Models","created":"2026-06-25","cited_by_count":1,"area":"safety"},{"id":"2607.01153","title":"Adversarial Pragmatics for AI Safety Evaluation: A Diagnostic Framework and Seed Benchmark for Language-Mediated Control","created":"2026-07-01","cited_by_count":1,"area":"agents"},{"id":"2607.12406","title":"Isolation as a First-Class Principle for LLM-Agent System Safety: Concepts, Taxonomy, Challenges and Future Directions","created":"2026-07-14","cited_by_count":1,"area":"agents"},{"id":"2608.02018","title":"Invisible Ink Threats: Adversarial Goals Behind Legitimate Tasks in Computer-Use Agents","created":"2026-08-03","cited_by_count":1,"area":"agents"},{"id":"2608.22248","title":"Beyond Over-Refusal: Defending Indirect Prompt Injection via Latent Instruction Manifolds","created":"2026-08-23","cited_by_count":1,"area":"safety"},{"id":"2601.11252","title":"Beyond Model Scaling: Test-Time Intervention for Efficient Deep Reasoning","created":"2026-01-16","cited_by_count":0,"area":"lr"},{"id":"2605.10763","title":"MATRA: Modeling the Attack Surface of Agentic AI Systems -- OpenClaw Case Study","created":"2026-05-11","cited_by_count":0,"area":"agents"},{"id":"2605.16471","title":"From AI-Generated Content to Agentic Action: Security and Safety Threats in Generative AI","created":"2026-05-15","cited_by_count":0,"area":"agents"},{"id":"2605.27784","title":"WIRE: Profiling Witnessed Within-Policy Instruction Collisions in LLM Agents","created":"2026-05-27","cited_by_count":0,"area":"agents"},{"id":"2605.28467","title":"Mitigating Adaptive Attacks against Reasoning Models with Activation Consistency Training","created":"2026-05-27","cited_by_count":0,"area":"lr"},{"id":"2606.29960","title":"IHDec: Divergence-Steered Contrastive Decoding for Securing Multi-Turn Instruction Hierarchies","created":"2026-06-29","cited_by_count":0,"area":null},{"id":"2608.09476","title":"ActBench: Self-Evolving Benchmark of Behavioral Safety in Cowork Agents","created":"2026-08-10","cited_by_count":0,"area":"agents"},{"id":"2608.23873","title":"Semantic Overlays: Mitigating Prompt Injection with Annotations Beyond Tokens and Steering Vectors","created":"2026-08-24","cited_by_count":0,"area":"safety"},{"id":"2609.10854","title":"No-Box Vulnerability Analysis: Description-only Detection of Indirect Prompt Injection Vulnerabilities in MCP Servers","created":"2026-09-09","cited_by_count":0,"area":"safety"},{"id":"2609.13725","title":"IBBench-Light: A Paired Evaluation of Task-Conditioned Responses to External Directives","created":"2026-09-12","cited_by_count":0,"area":null},{"id":"2609.16984","title":"Nameless Tokenization: A Lossless Tokenizer-Level Defense Against Control-Token Forgery in Open-Weight LLMs","created":"2026-09-15","cited_by_count":0,"area":null},{"id":"2609.20188","title":"ResumeShield: Channel Separation and an Open Benchmark for Indirect Prompt Injection in AI Resume Screening","created":"2026-07-24","cited_by_count":0,"area":"safety"},{"id":"2609.21088","title":"Origin Is All You Need: Provenance-Aware Transformers for Structural Trust-Boundary Separation","created":"2026-09-17","cited_by_count":0,"area":"safety"},{"id":"2609.23700","title":"When the Agent Becomes the Kernel: A Systematization of Security on the Path to AI-Native Operating Systems","created":"2026-09-20","cited_by_count":0,"area":"agents"},{"id":"2609.33179","title":"Scoring the Wrong Question: Readout Failures in Constrained-Option Evaluation","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.35932","title":"Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt Injection","created":"2026-09-28","cited_by_count":0,"area":"agents"},{"id":"2609.36570","title":"CounterSteer: Suppressing Indirect Prompt Injection with Activation Steering","created":"2026-09-29","cited_by_count":0,"area":"agents"},{"id":"2609.37196","title":"ToolFence: Fine-Grained Authorization for Secure Tool-Using LLM Agents","created":"2026-09-29","cited_by_count":0,"area":"agents"},{"id":"2610.01349","title":"PACE: Provenance-Aware Capability Enforcement for Tool-Using LLM Agents","created":"2026-10-01","cited_by_count":0,"area":"agents"},{"id":"2610.04210","title":"Can LLMs Separate Pasted Artifacts from User Speech? Absorption at Unmarked Prompt Seams","created":"2026-10-03","cited_by_count":0,"area":null},{"id":"2610.05139","title":"Hidden in the Comments: A Context-Injection Attack Surface in Code LLMs","created":"2026-10-04","cited_by_count":0,"area":null},{"id":"2610.18408","title":"Prompt-Injection Probes Learn the Conversation, Not Just the Injection","created":"2026-10-13","cited_by_count":0,"area":"safety"},{"id":"2610.18457","title":"Kairos: Transactional Security Enforcement at the System Boundary for LLM Agents","created":"2026-10-13","cited_by_count":0,"area":"agents"}],"next_cursor":null}