Security papers

Browse recent papers, open the full write-up, or filter the archive by topic.

62 papers found

ReviewPaperPublishedTopics

7.5BadRobot: Jailbreaking Embodied LLM Agents in the Physical WorldUnderstanding and mitigating the risks associated with embodied AI systems is crucial for ensuring their safe deployment in real-world applications.Jun 10, 2026LLM Agent Security

7.2Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and EvaluationAs LLM agents become more integrated into software systems, understanding their security vulnerabilities and defenses is crucial for safe deployment and operation.Jun 10, 2026LLM Agent Security

7.5Semantic Quorum Assurance: Collective Certification for Non-Deterministic AI InfrastructureAs AI systems become more integrated into critical infrastructure, ensuring their operational safety is paramount, making SQA a relevant contribution to the field.Jun 9, 2026Infra Security Safety

7.5Attack Selection in Agentic AI Control Evaluations Meaningfully Decreases SafetyUnderstanding the implications of strategic attack selection is crucial for developing more robust AI safety frameworks and ensuring accurate safety assessments.Jun 8, 2026Security Safety Evaluation

7.5Safety Paradox: How Enhanced Safety Awareness Leaves LLMs Vulnerable to Posterior AttackUnderstanding this safety paradox is crucial for developing more robust alignment mechanisms in LLMs, which is essential for their safe deployment in real-world applications.Jun 6, 2026LLM Alignment Security

7.5Will the Agent Recuse Itself? Measuring LLM-Agent Compliance with In-Band Access-Deny SignalsAs LLM agents take on more responsibilities, establishing mechanisms for compliance with access controls is crucial for safe and effective deployment in real-world applications.Jun 6, 2026LLM Agent Security

6.8Insurance of Agentic AIAs AI systems become more autonomous, understanding how to insure their risks is crucial for mitigating potential liabilities and ensuring safe deployment.Jun 6, 2026Agent Security Safety

7.5Toward Pre-Deployment Assurance for Enterprise AI Agents: Ontology-Grounded Simulation and Trust CertificationAs AI agents become integral to various industries, ensuring their compliance and safety before deployment is crucial for regulatory adherence and risk management.Jun 6, 2026Agent Security Safety

6.8Provably Auditable and Safe LLM Agents from Human-Authored OntologiesThis work is relevant for AI engineers and researchers interested in developing reliable and auditable LLM agents for complex domains, particularly in fields like healthcare and security.Jun 5, 2026LLM Agent Security

6.8When Safe Skills Collide: Measuring Compositional Risk in Agent Skill EcosystemsUnderstanding and mitigating compositional risks in AI systems is crucial for ensuring the safety and reliability of complex agentic AI deployments.Jun 2, 2026Agent Safety Security

6.8AI Loss of Control Incident Management: Response & ResilienceUnderstanding and managing AI loss of control incidents is crucial for ensuring safety and resilience in AI systems, especially as they become more autonomous and integrated into critical applications.Jun 1, 2026Safety Security

6.8AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and SecurityAI engineers and researchers should care because it addresses emerging safety risks in AI agents, offering a scalable solution that reduces deployment overhead and enhances real-time safety moderation.May 29, 2026Agent Safety Security

7.5Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent SystemsUnderstanding privacy risks in multi-agent systems is crucial for developing safer AI applications that operate in social environments.May 28, 2026LLM Agent Security

7.2Agent-ToM: Learning to Monitor Autonomous LLM Agents via Theory-of-Mind ReasoningUnderstanding and monitoring the intentions and beliefs of autonomous agents is crucial for ensuring their safe and reliable deployment in real-world applications.May 26, 2026LLM Agent Reasoning

7.5Methods for Formal Verification of Agent Skills: Three Layers Toward a Mechanically Checkable Capability-Containment ProofAs AI systems become more complex, ensuring the reliability and safety of agent skills is crucial for their deployment in real-world applications.May 26, 2026Agent Security

6.8Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system securityUnderstanding and improving the trustworthiness of agentic AI systems is crucial for their deployment in high-risk environments.May 26, 2026LLM Agent Security

7.2SafeHarbor: Hierarchical Memory-Augmented Guardrail for LLM Agent SafetyAs LLMs transition into autonomous agents, ensuring their safe operation in real-world environments is crucial to prevent misuse and maintain user trust.May 25, 2026LLM Agent Safety

6.8Measuring Safety Alignment Effects in Autonomous Security AgentsUnderstanding safety alignment in autonomous agents is crucial for developing reliable security systems that can effectively handle vulnerabilities without compromising safety.May 21, 2026Agent Security Safety

6.8The Insurability Frontier of AI Risk: Mapping Threats to Affirmative Coverage, Silent Exposures, and ExclusionsUnderstanding the insurability of AI risks is crucial for AI engineers and researchers to navigate potential liabilities and ensure adequate coverage in an increasingly AI-driven world.May 21, 2026Security Safety Survey

7.2AgentWall: A Runtime Safety Layer for Local AI AgentsAs AI agents gain more capabilities, ensuring their safe operation in local environments is crucial to prevent unintended or malicious actions, making AgentWall relevant for developers and researchers working with autonomous systems.May 19, 2026Agent Safety Security

6.5AI Consciousness and Existential RiskUnderstanding the nuances between consciousness and intelligence in AI is crucial for developing effective safety measures and policies to mitigate existential risks.May 18, 2026Alignment Security Safety

7.8Autonomous Intelligent Agents for Natural-Language-Driven Web Execution with Integrated Security AssuranceThis work addresses critical challenges in web testing and security, providing a practical solution that can save developers time and improve software reliability.May 18, 2026Agent Security Coding

7.2Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDPUnderstanding cost-effective design choices for LLM agents in complex environments can guide practitioners in developing more efficient and effective AI systems.May 18, 2026LLM Agent Reasoning

7.5MemLineage: Lineage-Guided Enforcement for LLM Agent MemoryAs LLMs become more integrated into critical applications, ensuring the integrity of their memory and preventing malicious influences is crucial for safety and reliability.May 16, 2026LLM Agent Security

7.2Towards Security-Auditable LLM Agents: A Unified Graph RepresentationAs LLM-based systems become more complex and autonomous, ensuring their security through effective auditing mechanisms is crucial for preventing and understanding security breaches.May 11, 2026LLM Agent Security

6.8When Safety Geometry Collapses: Fine-Tuning Vulnerabilities in Agentic Guard ModelsUnderstanding and mitigating safety vulnerabilities in AI guard models is crucial for ensuring reliable and secure deployment of agentic AI systems.May 7, 2026Safety Alignment Security

7.0Safety in Embodied AI: A Survey of Risks, Attacks, and DefensesUnderstanding safety in embodied AI is crucial as these systems are increasingly deployed in safety-critical environments like healthcare and transportation.May 7, 2026Survey Safety Security

6.8Can LLMs Make (Personalized) Access Control Decisions?Understanding how LLMs can automate access control decisions could reduce user burden and improve security in complex systems.May 7, 2026LLM Security Safety

6.8Multi-Agent Strategic Games with LLMsUnderstanding how LLMs can simulate strategic interactions offers a novel approach to studying complex social dynamics and international relations theories.May 7, 2026LLM Agent Reasoning

7.5Stable Agentic Control: Tool-Mediated LLM Architecture for Autonomous Cyber DefenseThis research addresses critical needs in cybersecurity by providing a framework that ensures reliable decision-making in high-stakes environments, which is essential for modern security operations.May 7, 2026LLM Agent Security

6.8AI Safety as Control of Irreversibility: A Systems Framework for Decision-Energy and Sovereignty BoundariesUnderstanding and managing the potential for AI systems to make irreversible decisions is crucial for ensuring they remain under human control and do not inadvertently cause systemic harm.May 6, 2026Safety Alignment Security

7.2Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety AlignmentThis research is relevant for improving the robustness of language models in high-risk scenarios by mitigating vulnerabilities to persona-based attacks.May 6, 2026LLM Alignment Safety

8.2When Alignment Isn't Enough: Response-Path Attacks on LLM AgentsUnderstanding and mitigating response-path attacks is crucial for ensuring the reliability and security of LLM applications, especially in sensitive contexts.May 6, 2026LLM Agent Security

7.2Semia: Auditing Agent Skills via Constraint-Guided Representation SynthesisUnderstanding and mitigating security risks in agent skills is crucial for ensuring the safe deployment of AI-driven systems, making this research relevant for AI engineers and researchers focused on security.May 5, 2026LLM Agent Security

7.5Attention Is Where You AttackUnderstanding and mitigating adversarial attacks like ARA is crucial for improving the safety and robustness of large language models in real-world applications.May 5, 2026LLM Alignment Security

7.2IACDM: Interactive Adversarial Convergence Development Methodology -- A Structured Framework for AI-Assisted Software DevelopmentAs AI tools become more prevalent in software development, understanding and addressing their limitations is crucial for ensuring the reliability and security of AI-generated applications.May 1, 2026Coding Security Evaluation

7.5Knowledge Graph Representations for LLM-Based Policy Compliance ReasoningAs AI regulations become more critical, this research provides a method to improve compliance reasoning, which is essential for developing safe AI applications.May 1, 2026LLM Reasoning Security

7.5Risk Reporting for Developers' Internal AI Model UseAs AI models become more capable, understanding and managing the risks associated with their internal deployment is crucial for safety and compliance.Apr 30, 2026Security Safety Evaluation

6.8Data-Centric Foundation Models in Computational Healthcare: A SurveyUnderstanding data-centric approaches in foundation models can improve healthcare AI systems by addressing data quality and ethical challenges.Apr 30, 2026Survey Data Alignment

6.8The Kerimov-Alekberli Model: An Information-Geometric Framework for Real-Time System StabilityThis work offers a potentially rigorous approach to AI safety by grounding ethical violations in physical and informational principles, which could lead to more reliable autonomous systems.Apr 29, 2026Safety Alignment Security

7.2AI Identity: Standards, Gaps, and Research Directions for AI AgentsAs AI agents increasingly operate autonomously, understanding their identity and accountability is crucial for safe and effective deployment in real-world applications.Apr 29, 2026Agent Survey Security

7.0AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and SecurityAI engineers and researchers should care because AgentDoG offers a structured approach to diagnosing and mitigating complex safety and security risks in AI agents, which is crucial for deploying reliable autonomous systems.Apr 24, 2026Agent Safety Security

7.2LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM SafetyImproving LLM safety across languages is crucial for deploying these models globally without compromising security in low-resource languages.Apr 24, 2026LLM Alignment Safety

6.8Reinforcing privacy reasoning in LLMs via normative simulacra from fictionImproving LLMs' understanding of privacy norms is crucial for their safe deployment in real-world applications where privacy expectations vary by context.Apr 24, 2026LLM Reasoning Alignment

7.3AI Governance under Political Turnover: The Alignment Surface of Compliance DesignUnderstanding the implications of AI governance in political contexts is crucial for ensuring that automated systems remain accountable and do not become tools for manipulation.Apr 24, 2026Alignment Security Other

7.5AVISE: Framework for Evaluating the Security of AI SystemsAs AI systems become integral to critical applications, understanding and mitigating their security vulnerabilities is essential for ensuring their safe deployment.Apr 23, 2026Security Evaluation

7.5Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture The Flag ChallengesUnderstanding the limitations of LLM agents in cybersecurity tasks is crucial for developing more effective autonomous systems in this domain.Apr 22, 2026LLM Agent Evaluation

7.5Owner-Harm: A Missing Threat Model for AI Agent SafetyUnderstanding and mitigating owner-harm is crucial for the safe deployment of AI agents, especially as their use becomes more prevalent in sensitive commercial environments.Apr 22, 2026Security Safety Agent

7.5The Consensus Trap: Rescuing Multi-Agent LLMs from Adversarial Majorities via Token-Level CollaborationAs multi-agent systems become more prevalent, ensuring their resilience against adversarial attacks is crucial for reliable AI applications.Apr 21, 2026LLM Agent Reasoning

7.5The Blind Spot of Agent Safety: How Benign User Instructions Expose Critical Vulnerabilities in Computer-Use AgentsUnderstanding these vulnerabilities is crucial for developing safer AI systems, especially as CUAs become more prevalent in complex digital environments.Apr 21, 2026Other