Alignment papers

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

133 papers found

ReviewPaperPublishedTopics

7.5LLMs Can Better Capture Human Judgments--With the Right PromptsImproving LLMs' alignment with human judgments is crucial for applications in ethics, decision-making, and social sciences, where understanding human perspectives is essential.Jun 12, 2026LLM Alignment Evaluation

6.8Definitional alignment before capability alignment: a Design-Science framework for adjudicating claims about AGIThis work is relevant for researchers and engineers as it provides a structured approach to navigate the complex landscape of AGI definitions, which is crucial for effective communication and governance in AI development.Jun 12, 2026Alignment Evaluation Other

6.5Existential Indifference: Self-Nonpreservation as a Necessary Architectural Condition for Aligned Superintelligence (or: The Suicidal AI)Understanding and addressing the root causes of AI misalignment is crucial for developing safe and aligned superintelligent systems.Jun 11, 2026Alignment Safety

7.5From Consumption to Reflection: Designing Human-AI Relations for Stable ReasoningThis work addresses critical cognitive limitations in AI interactions, potentially improving decision-making processes in applications reliant on LLMs.Jun 11, 2026LLM Reasoning Alignment

6.8Certifiable Safe RLHF: Semantic Grounding and Fixed Penalty Constraint Optimization for Safer LLM AlignmentEnsuring the safety of LLMs is crucial for their deployment in real-world applications, and this paper proposes a method that could improve safety without the computational overhead of traditional approaches.Jun 11, 2026LLM Alignment Safety

7.5ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit MixingThis approach addresses a critical safety issue in large language models, allowing for improved robustness against harmful prompts while maintaining task performance.Jun 11, 2026Alignment Safety LLM

6.8An LLM-Native Psychometric Instrument Does Not Predict LLM Behavior: Evidence Across 25 ModelsUnderstanding the limitations of LLM self-reports is crucial for improving model alignment and reliability in real-world applications.Jun 10, 2026LLM Alignment Evaluation

6.8Emergent alignment and the projectability of ethical personasUnderstanding how language models can be aligned with ethical frameworks is crucial for developing AI systems that behave predictably and ethically in diverse scenarios.Jun 10, 2026LLM Alignment Safety

6.8The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent MisalignmentAs AI systems increasingly rely on multi-agent interactions, ensuring alignment among agents is crucial for reliable decision-making and task execution.Jun 10, 2026Agent Alignment Evaluation

7.5Impacts of Histories and Models on LLM Grading: A Study in Advanced Software Engineering CoursesUnderstanding LLM grading reliability is crucial for educators considering automation in assessment, as it directly impacts educational fairness and workload.Jun 9, 2026LLM Evaluation Alignment

6.5Position: Anthropomorphic Misalignment Research Needs Stronger EvidenceAI engineers and researchers should care because robust evidence in AMR is crucial for making informed safety decisions regarding AI deployment and regulation.Jun 9, 2026Alignment Safety Evaluation

6.8Symbolic Reasoning Frameworks Modulate LLM Risk Aversion in Multi-Agent Strategic SettingsUnderstanding how symbolic reasoning frameworks affect language model behavior in strategic settings could inform the design of more effective AI agents in complex environments.Jun 9, 2026LLM Agent Reasoning

6.8LCAM: A Framework for Diagnosing Interactional Alignment Failures in Con-versational AIUnderstanding and diagnosing interactional failures in conversational AI is crucial for improving user trust and safety, especially in sensitive applications like counseling.Jun 9, 2026Alignment Evaluation

6.8Modality Gap-Driven Subspace Alignment Training Paradigm For Multimodal Large Language ModelsThis research could reduce the dependency on expensive image-text datasets, making it easier and more cost-effective to train multimodal models at scale.Jun 8, 2026Multimodal Vision Language Alignment

7.5MVCL-DAF++: Enhancing Multimodal Intent Recognition via Prototype-Aware Contrastive Alignment and Coarse-to-Fine Dynamic Attention FusionThis research addresses critical challenges in multimodal intent recognition, making it relevant for applications requiring robust understanding in noisy environments.Jun 8, 2026Multimodal Alignment

7.5Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision ContextsUnderstanding process alignment can lead to more accurate and ethically sound AI systems that better reflect organizational values and decision-making processes.Jun 6, 2026Alignment Evaluation

7.5Toward Culturally Aligned LLMs through Ontology-Guided Multi-Agent ReasoningThis research addresses the critical issue of cultural misalignment in LLMs, which is essential for developing AI systems that are sensitive to diverse cultural values and improve decision-making processes.Jun 6, 2026LLM Agent Reasoning

7.5Trust, but Don't Verify: Epistemic Blind Spots in LLM Source EvaluationUnderstanding the limitations of language models in evidence evaluation is crucial for improving their reliability in decision-making contexts.Jun 6, 2026LLM Evaluation Alignment

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

6.5Towards AI epidemiology: a measurement standardisation framework for prospective risk detectionThis framework could enhance the reliability and governance of AI systems, providing a basis for monitoring and improving alignment in various applications.Jun 6, 2026Alignment Evaluation

7.5Where Should Knowledge Enter? A Layered Framework for Knowledge Infusion in Multimodal Iterative Generative MoThis research addresses the critical challenge of ensuring reliability in generative models when handling domain-specific knowledge, which is essential for applications in safety-critical areas.Jun 6, 2026Multimodal Alignment Safety

6.8Inference-Time Vulnerability Beyond Shallow Safety: Alignment Along Generation TrajectoriesUnderstanding and mitigating inference-time vulnerabilities is crucial for deploying safer and more reliable language models in real-world applications.Jun 5, 2026LLM Alignment Safety

7.5BioBlue: Systematic runaway-optimiser-like LLM failure modes on biologically and economically aligned AI safety benchmarks for LLMs with simplified observation formatUnderstanding these failure modes is crucial for developing safer AI systems, particularly as LLMs are increasingly deployed in complex decision-making scenarios.Jun 5, 2026LLM Alignment Safety

6.8RUBAS: Rubric-Based Reinforcement Learning for Agent SafetyAs AI agents increasingly interact with real-world tools, ensuring their safe operation is crucial for preventing unintended consequences and maintaining user trust.Jun 5, 2026LLM Agent Safety

7.5The Digital Apprentice: A Framework for Human-Directed Agentic AI DevelopmentThis framework addresses the critical challenge of ensuring accountability in AI systems while enabling their scalability, which is essential for practical applications in various fields.Jun 5, 2026Agent Alignment Architecture

7.5Efficiently Aligning Language Models with Online Natural Language FeedbackThis research addresses the challenge of training language models in complex domains with limited expert input, which is crucial for practical AI applications.Jun 5, 2026Alignment LLM

6.8Fully Automated Identification of Lexical Alignment and Preference-Stage Shifts in Large Language ModelsUnderstanding and improving lexical alignment in language models can enhance their interaction quality and alignment with human expectations, which is crucial for developing more effective AI communication tools.Jun 3, 2026LLM Alignment Evaluation

7.5Mechanism Design Is Not Enough: Prosocial Agents for Cooperative AIThis research highlights the limitations of existing mechanisms in AI safety and proposes a novel approach that could lead to more effective cooperative interactions among AI systems.Jun 3, 2026Agent Alignment Safety

7.5Decomposing and Measuring Evaluation AwarenessUnderstanding evaluation awareness is crucial for improving the validity of benchmark results and ensuring that language models behave consistently across different evaluation contexts.Jun 3, 2026Evaluation Benchmark Alignment

7.2SafeSteer: Localized On-Policy Distillation for Efficient Safety AlignmentThis work is relevant for AI engineers and researchers seeking efficient methods to align language models with human values without significantly degrading their general capabilities.Jun 2, 2026LLM Alignment Safety

6.8ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer UseUnderstanding and addressing AI misalignment in everyday settings is crucial for ensuring the safe deployment of autonomous agents in real-world applications.Jun 2, 2026Agent Safety Alignment

7.5Isolating LLM Lexical Bias: A Curation-Free Triangulated Metric for Preference-Stage LearningUnderstanding and quantifying lexical bias is crucial for improving model alignment and ensuring the trustworthiness of AI systems.Jun 2, 2026LLM Alignment Evaluation

7.5Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMsThis work could significantly enhance the performance of reasoning LLMs in diverse tasks, making them more robust and adaptable in real-world applications.Jun 2, 2026LLM Reasoning Alignment

7.5MESA: Improving MoE Safety Alignment via Decentralized ExpertiseAs large language models become more prevalent, ensuring their safety and alignment is crucial to prevent misuse and enhance their reliability in real-world applications.Jun 2, 2026Alignment Safety Architecture

6.8EUDAIMONIA: Evaluating Undesirable Dynamics in AIUnderstanding and mitigating the social risks of LLMs is crucial for ensuring user welfare and ethical AI deployment in real-world applications.Jun 1, 2026LLM Alignment Evaluation

6.8COMPASS: Cognitive MCTS-Guided Process Alignment for Safe Search AgentsEnsuring the safety of AI systems, especially those capable of multi-step reasoning, is crucial for preventing harmful outcomes and maintaining trust in AI technologies.Jun 1, 2026LLM Agent Safety

7.2Multi-Agent Teams Hold Experts BackUnderstanding the limitations of multi-agent systems in leveraging expertise is crucial for improving collaborative AI applications in real-world scenarios.Jun 1, 2026LLM Agent Alignment

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

7.2The Alignment Floor: When Persona Customization Is SafeUnderstanding the limits of persona customization in AI models is crucial for ensuring safe and effective deployment in diverse applications, particularly in maintaining alignment and reducing sycophancy.May 28, 2026Alignment Evaluation Safety

7.5AI, Take the Wheel: What Drives Delegation and Trust in Human-Computer Cooperative Question Answering?Understanding how humans decide to trust AI is crucial for designing better collaborative systems that enhance performance and user experience.May 28, 2026Agent Alignment Evaluation

7.2SafeMed-R1: Clinician-Audited Safety and Ethics Alignment for Medical Large Language ModelsThis work is relevant for AI researchers and engineers focused on deploying LLMs in healthcare, as it addresses critical safety and ethical concerns that are barriers to clinical adoption.May 28, 2026LLM Safety Alignment

7.2Modeling Community Attitude through Reaction Tone: A Human-AI Collaborative Framework for Evaluating LLM Alignment with Linguistic Behaviors in Online CommunitiesUnderstanding the limitations of LLMs in simulating community attitudes is crucial for improving their application in social analysis and ensuring more accurate representations of human behavior.May 28, 2026LLM Alignment Evaluation

6.8Position: AI Safety Requires Effective ControllabilityEnsuring AI systems can be controlled and overridden in real-time is crucial for deploying them safely in dynamic environments.May 27, 2026Safety Alignment Benchmark

6.8Curriculum Learning for Safety AlignmentImproving the robustness of safety alignment in language models is crucial for deploying them in real-world applications where safety and reliability are paramount.May 27, 2026LLM Alignment Safety

6.8LURE: Live-Usage Replay Evaluations for Reducing Evaluation AwarenessUnderstanding and reducing evaluation awareness in language models is crucial for developing reliable safety and alignment benchmarks, which are essential for deploying AI systems in real-world scenarios.May 27, 2026LLM Evaluation Alignment

7.2Think Twice Before You Act: Enhancing Agent Behavioral Safety with Thought CorrectionThis approach provides a scalable solution to improve the safety of AI agents without altering their underlying models, which is crucial for deploying AI in real-world applications where safety is paramount.May 27, 2026LLM Agent Safety

6.8ALIVE: Awakening LLM Reasoning via Adversarial Learning and Instructive Verbal EvaluationThis approach could lead to more autonomous and adaptable AI systems by reducing reliance on costly and brittle reward signals.May 25, 2026LLM Reasoning Alignment

7.5EvalMORAAL: Interpretable Chain-of-Thought and LLM-as-Judge Evaluation for Moral Alignment in Large Language ModelsUnderstanding moral alignment in AI is crucial for developing culturally aware systems that can operate effectively across diverse global contexts.May 22, 2026Alignment Evaluation Interpretability

6.8Towards Context-Invariant Safety Alignment for Large Language ModelsThis approach could lead to more reliable and robust language models that maintain safety across varied and potentially adversarial contexts, which is crucial for deploying LLMs in real-world applications.May 22, 2026LLM Alignment Safety

6.8Distributional AGI SafetyUnderstanding and mitigating risks associated with distributed AGI systems is crucial as AI systems become more interconnected and capable of coordination.May 21, 2026Safety Alignment Agent