Latest AI Safety & Alignment Research Papers

The newest AI Safety & Alignment papers from across the field — arXiv, NeurIPS, CVPR, Nature, and more — refreshed daily and ranked by relevance. Distill AI tracks AI Safety & Alignment so you don’t have to: get the standout work delivered to your inbox every morning, with 2-sentence summaries and the option to chat with any paper.

Recent papers

• Multi-dimensional hierarchical temporal alignment for improved temporal commonsense reasoning in large language models
  Ge Yan, Hai-Tao Yu, Lei Chao, 30751052 et al. · Institutional Repositories ... · Nov 1, 2026

  Benefiting from recent advances in generative AI and large language models (LLMs), current LLM-driven AI systems are rapidly reshaping how we reason, plan, and make decisions. Yet endowing these systems with human-like temporal intelligence…

• How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates
  Max Neuwinger, Hanna Yukhymenko, Imanol Schlag · arXiv · Oct 29, 2026

  Safety alignment for Large Language Models is heavily focused on English with limited multilingual representation, particularly damaging low-resource language performance. Related open training datasets are often limited in size or do not f…

• Learned Terminal Costs for Short-Horizon Model Predictive Control of Autonomous Trucks in Interactive Highway Traffic
  Hooman Hamze, Benjamin Houshmand, Deepthi Pathare, Erik Börve et al. · arXiv · Oct 29, 2026

  Autonomous highway driving for heavy-duty trucks is a challenging control problem. The greater mass, dimensions and intricate dynamics increase the requirements for cooperation from surrounding traffic in critical maneuvers, requiring a car…

• Learning Safe Flow Matching Robot Policies from Binary Feasibility Oracles
  Jianming Ma, Rongjun Jin, Yizhi Chen, Yue Gao · arXiv · Oct 29, 2026

  Robot foundation models increasingly rely on flow-based action generation for general-purpose manipulation, yet the generated actions are not guaranteed to satisfy physical constraints such as joint limits and collision avoidance. This prob…

• Towards sustainable construction: A Roadmap for Aligning BES6001 with the UN SDGs
  Jacqui Glass, Abdul‐Majeed Mahamadu, Sophie Ball, Colin Booth et al. · UCL Discovery (University C... · Oct 29, 2026

  The architecture, engineering and construction (AEC) sector has grown significantly over recent decades, sparking ongoing debate about its environmental impact. While advancing sustainability is essential, the sector’s role in meeting housi…

• Timestral
  Adrien Petralia, Alexandre Gavaudan, Artjom Joosen, Etienne Goffinet et al. · arXiv · Oct 27, 2026

  We introduce Timestral, a family of four time series foundation models for probabilistic forecasting, with 5M, 28M, 123M, and 209M parameters. Timestral sets a new state of the art on GIFT-Eval, TIME, and fev-bench, where Timestral-209M and…

• When uncertainty quantification fails: structural vulnerabilities and corrections for selective prediction under clinical class imbalance
  Walter Endrizzi, Flavio Ragni, Raffaele Marchesi, Stefano Bovo et al. · arXiv · Oct 27, 2026

  Background. Selective prediction systems improve the safety of clinical artificial intelligence by deferring ambiguous cases to human experts while automating high-confidence decisions. However, this safety mechanism relies entirely on the …

• Follow the Gradient: Tracing Safety Degradation Back to Training Data
  Saket Reddy, Amro Abbas, Anshuman Suri, Ari Morcos et al. · arXiv · Oct 27, 2026

  Fine-tuning Large Language Models (LLMs) on instruction data can quietly erode their safety, even when no training example looks harmful. Because this safety-degrading data looks benign, standard content filters struggle to catch it. We tac…

• M4Bench: Evaluating Procedural Specification for Clinical EHR Derivation Agents
  Hannes Ill, Rafi Al Attrach, Rajna Fani, Ahram Han et al. · arXiv · Oct 27, 2026

  Clinical research over electronic health records depends on derived variables whose definitions sit outside the schema, in item codes, time windows, missingness rules, and output-grain choices that change cohorts and downstream conclusions.…

• Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
  Zirui Ding, Xinxin Fan, Hao Yuan, Shaoye Luo et al. · arXiv · Oct 27, 2026

  Large language models (LLMs) are usually evaluated by final-output correctness or safety, yet an endpoint label does not reveal how a failure develops throughout the reasoning process. We therefore ask whether ordinary reasoning failures an…

• Aligning Institutional Power and Operational Ability
  Andrea Marino, Giovanni Sileno, Thomas van Binsbergen, Tom van Engers · arXiv · Oct 27, 2026

  To be socially sustainable, Agentic AI systems require normative designs that align responsibilities with the ability to fulfil them. Identifying incongruences such as unsatisfiable expectations or privilege escalation is crucial in avoidin…

• Aligning Large Language Models for Safety in African Languages: A Case Study of Hausa
  Valdini Douglace Lemofouet, Blessing Ngozi Uzor, Paula Chikaodinaka Anyanwu, Danielle Blanche Kapsa et al. · arXiv · Oct 27, 2026

  Safety alignment research for large language models (LLMs) remains concentrated on high-resource languages, leaving open how existing alignment methods perform in low-resource, culturally distinct settings. We study this question for Hausa,…

• I-HITL: Inalienable Human-in-the-Loop
  Samuel Breckenridge, Dani Vilardell, Wenhao Wang, Fan Zhang et al. · arXiv · Oct 26, 2026

  Human oversight is a pillar of AI safety. It is often exercised by a human in the loop (HITL) who approves critical actions taken by an AI agent. AI agents' increasing ability to bypass guardrails, though, raises the question of how to enfo…

• Multi-Objective Reinforcement Learning Using Routed Ensembles and Trajectory Attribution
  Zhenshuo Zhang, Ruoxuan Xiong, Hongyang R. Zhang · arXiv · Oct 26, 2026

  Multi-objective reinforcement learning is a fundamental problem for modern AI, with applications spanning post-training, reasoning models, and robotic control. A central challenge is to learn policies that balance competing objectives, such…

• The Watchtower Imperative: Lifespan-Calibrated Evaluation for Frontier AI
  Casey M. Williams, Jennifer Victoria Scurrell · arXiv · Oct 26, 2026

  The gap between AI capability deployment and safety assurance is structural and cannot be closed through internal evaluations alone. Analyzing documented harms (2017-2026) across the human lifespan reveals that disparate incidents --- from …

• From Language to Drives: A Multimodal Framework for Aligning Autonomous Robots with Human Purposes
  Diana Cerdas-Vargas, Alejandro Romero, Richard J. Duro · arXiv · Oct 26, 2026

  This work explores a multimodal approach to the alignment of the behavior of open-ended learning autonomous robots with human purposes through the use of large language models as cultural enablers. While autonomous robots can generate their…

• Interpretable sparse fine-tuning with TopKLoRA enables faithful circuit discovery
  Marek Masiak, Andrzej Szablewski, Nicola Cancedda, Constantin Venhoff · arXiv · Oct 23, 2026

  Fine-tuning on datasets of unknown origin can create a backdoor in the trained model when the data includes poisoned examples. Difficult to localise and remove, backdoors can persist through safety training. While parameter-efficient fine-t…

• Clean-label Poisoning Attack and Defense Game for Federated Learning in Vehicular Networks
  Yalin E. Sagduyu, Tugba Erpek, Nathan Stephenson, Mengran Xue · arXiv · Oct 23, 2026

  This paper studies interactions between clean-label poisoning attack and defense in vehicular networks using federated learning (FL). Basic Safety Message (BSM) classification for anomaly detection is modeled as a distributed learning probl…

• SP-Guide: Enhancing Aerial Vision-Language Navigation with Spatial Priors
  Haoxiang Tang, Dong Zhou, Shen Tan, Minxuan Jiang et al. · arXiv · Oct 23, 2026

  Aerial vision-language navigation (AVLN) requires a UAV to reach a language-described target from partial egocentric observations while reasoning over large outdoor scenes. Existing approaches typically fuse view-centered RGB observations a…

• ALLMTS: Automated LLM-based Medical Triage System
  Michał Remigiusz Janiszewski, Artur Stopa, Łukasz Lenkiewicz, Wojciech Achtelik et al. · arXiv · Oct 23, 2026

  Telephone triage is a safety critical task: from a brief intake note, a nurse must decide whether a caller needs an ambulance, an emergency department, an appointment, or self care at home. Queried directly, large language models (LLMs) are…

• FOCUS: A Lightweight Mamba-2 Model for EEG-Based Cognitive Workload Classification
  Ioannis Kyprakis, Stefanos Gkikas, Chariklia Chatzaki, Zongyi Jiang et al. · arXiv · Oct 23, 2026

  Cognitive workload estimation from physiological signals is increasingly relevant in high-risk environments, where continuous and unobtrusive monitoring can inform adaptive systems and safety interventions. This paper presents FOCUS (Freque…

• EpiSEM-Bench: Evaluating Safety and Evidence Recovery in LLM Agents under Missing Evidence
  Qijin Zhang, Hanzhen Wang, Yuehan Zhao, Haoyu Dai et al. · arXiv · Oct 22, 2026

  Tool-using language model agents rely on environmental facts to decide whether an action is permitted. When a tool response omits a safety-relevant fact, an agent may carry out a prohibited action even though the user request is benign and …

• Simulated Control Evals
  Jeffrey Olmo, David Zhang, Tyler Tracy · arXiv · Oct 22, 2026

  Control evaluations measure whether safeguards can prevent an AI model from causing undesired outcomes even when it is intentionally trying to do so. These evaluations typically use purpose-built control settings and extrapolate their resul…

• Certified Execution Harness for Agentic AI: Safe and Least-Intrusive Petri-Net Supervisory Control Between Reasoning and Action
  Chao Gu, Quanyan Zhu · arXiv · Oct 22, 2026

  This paper studies the problem of safe tool execution for large language model (LLM) agents under prompt injection. Using Petri nets, we formalize the execution harness between the LLM and its tools as a supervisory control problem, in whic…

• MQAScore: Fine-Grained Text-Music Alignment Evaluation via Large Audio Language Model
  Wei-Jaw Lee, Chun-Ping Wang, Xuanjun Chen, Yi-Hsuan Yang · arXiv · Oct 22, 2026

  Text-to-music alignment is commonly evaluated by the global similarity between text and audio embeddings, which collapses all described attributes into a single score and obscures which ones are missing. We instead introduce MQAScore, which…

• Harm Is Behavioural, Moderation Is Lexical: How Language and Culture Decide Whom Automated Safety Systems Protect and Whom They Silence
  Gabriel Lars Sabadin, Dr Nicola Harding, Sarah Kingston · arXiv · Oct 22, 2026

  Content warning: this paper quotes slurs and hateful statements from published test suites; the racial slur among them is masked. Automated systems decide at scale which messages are harmful, and they decide unevenly. Classifiers label Afri…

• Differentially Private Sparse Reward Estimation with Preference Feedback
  Meng Ding, Mingxi Lei, Jie Zhang, Jinyan Liu et al. · arXiv · Oct 21, 2026

  Reinforcement Learning with Human Feedback (RLHF) has become a central paradigm for aligning Large Language Models with human values. However, the reliance on sensitive preference data necessitates rigorous differential privacy (DP) guarant…

• COMET: A coordinated multi-layer runtime monitoring framework for human-supervised AI-enabled safety systems
  Niclas Flehmig, Mary Ann Lundteigen, Shen Yin · arXiv · Oct 21, 2026

  Incorporating AI systems into safety systems requires monitoring to avoid silent faults and ensure safe and reliable operation. Single monitors often do not provide actionable information for human supervisors, who act as the system's adapt…

• Humans Correct Their Judgments Through Debate, but Weaker AI Models Do Not
  Maria Victoria Carro, Denise Alejandra Mester, Facundo Nieto, Oscar Agustin Stanchi et al. · arXiv · Oct 20, 2026

  How can we effectively oversee AI systems that surpass human intelligence? AI Safety via Debate proposes that two AI systems argue opposing positions and a judge, either a human or a weaker model, decides which is correct. By facilitating s…

• CoopUQ: Robust Multi-Robot Social Navigation via Uncertainty-Aware Cooperative Forecasting
  Zehao Wang, Jianpeng Yao, Yuping Wang, Yiping Li et al. · arXiv · Oct 19, 2026

  Reinforcement learning enables mobile robots to make fast and scalable navigation decisions in constrained human-shared environments. However, directly extending single-robot approaches to multi-robot settings often fails to ensure safety a…