safety

Memorization In Stable Diffusion Is Unexpectedly Driven by CLIP Embeddings

arXiv:2605.02908v1 Announce Type: new Abstract: Understanding how textual embeddings contribute to memorization in text-to-image diffusion models is crucial for both interpretability and safety. This

MILD: Mediator Agent System with Bidirectional Perception and Multi-Layered Alignment for Human-Vehicle Collaboration

DGX agent

arXiv:2605.01507v1 Announce Type: new Abstract: Prior studies report that partial driving automation can increase the cognitive demands on human drivers. This effect largely arises from human drivers'

safetyarxiv-cs-ai

6 May 2026

MINT: Minimal Information Neuro-Symbolic Tree for Objective-Driven Knowledge-Gap Reasoning and Active Elicitation

arXiv:2602.05048v2 Announce Type: replace Abstract: Joint planning through language-based interactions is a key area of human-AI teaming. Planning problems in the open world often involve various aspe

Mira Murati tells the court that she couldn’t trust Sam Altman’s words

Mira Murati, OpenAI's former CTO, has testified under oath that CEO Sam Altman lied to her about the safety standards for a new AI model. In a video deposition shown during the ongoing Musk v. Altman

safetythe-verge-ai

Mira Murati’s testimony is gripping – and what it makes absolutely clear is how utterly wrong most of Twitter was about why Sam was fired. –…

Mira Murati’s testimony is gripping – and what it makes absolutely clear is how utterly wrong most of Twitter was about why Sam was fired. – It had nothing per se to do with AI safety - It had nothing

safetygary-marcus--x

Mix3R: Mixing Feed-forward Reconstruction and Generative 3D Priors for Joint Multi-view Aligned 3D Reconstruction and Pose Estimation

arXiv:2605.03359v1 Announce Type: new Abstract: Recent trends in sparse-view 3D reconstruction have taken two different paths: feed-forward reconstruction that predicts pixel-aligned point maps withou

Model Routing as a Trust Problem: Route Receipts for Adaptive AI Systems

arXiv:2605.01710v1 Announce Type: new Abstract: AI products often route requests through version aliases, service tiers, tool choices, regional endpoints, fallback rules, or safety handling before res

Model Spec Midtraining: Improving How Alignment Training Generalizes

arXiv:2605.02087v1 Announce Type: new Abstract: Some frontier AI developers aim to align language models to a Model Spec or Constitution that describes the intended model behavior. However, standard a

Multilingual Safety Alignment via Self-Distillation

arXiv:2605.02971v1 Announce Type: cross Abstract: Large language models (LLMs) exhibit severe multilingual safety misalignment: they possess strong safeguards in high-resource languages but remain hig

safetyarxiv-cs-cl

Musk v. Altman: Mira Murati testifies that Sam Altman lied to her about the safety standards for a new OpenAI model and that he made her work more difficult (Jay Peters/The Verge)

Jay Peters / The Verge: Musk v. Altman: Mira Murati testifies that Sam Altman lied to her about the safety standards for a new OpenAI model and that he made her work more difficult — OpenAI's former C

safetytechmeme

Neuron-Anchored Rule Extraction for Large Language Models via Contrastive Hierarchical Ablation

arXiv:2605.03058v1 Announce Type: new Abstract: A key goal of explainable AI (XAI) is to express the decision logic of large language models (LLMs) in symbolic form and link it to internal mechanisms.

NORA: A Harness-Engineered Autonomous Research Agent for End-to-End Spatial Data Science

arXiv:2605.02092v1 Announce Type: new Abstract: The automation of scientific research workflows has emerged as a transformative frontier in artificial intelligence, yet existing autonomous research ag

Nora: Normalized Orthogonal Row Alignment for Scalable Matrix Optimizer

arXiv:2605.03769v1 Announce Type: new Abstract: Matrix-based optimizers have demonstrated immense potential in training Large Language Models (LLMs), however, designing an ideal optimizer remains a fo

Normalized Matching Transformer

arXiv:2503.17715v3 Announce Type: replace Abstract: We introduce the Normalized Matching Transformer (NMT), a deep learning approach for efficient and accurate sparse semantic keypoint matching betwee

OGPO: Sample Efficient Full-Finetuning of Generative Control Policies

arXiv:2605.03065v1 Announce Type: new Abstract: Generative control policies (GCPs), such as diffusion- and flow-based control policies, have emerged as effective parameterizations for robot learning.

On Surprising Effects of Risk-Aware Domain Randomization for Contact-Rich Sampling-based Predictive Control

arXiv:2605.03290v1 Announce Type: new Abstract: Domain randomization (DR) is widely used in policy learning to improve robustness to modeling error, but remains underexplored in contact-rich sampling-

OpenAI violated Canadian privacy laws in developing first ChatGPT model, probe finds https://www.theglobeandmail.com/business/article-openai…

OpenAI violated Canadian privacy laws in developing first ChatGPT model, probe finds https://www.theglobeandmail.com/business/article-openai-chatgpt-violated-canadian-privacy-laws-watchdogs-report/?ut

safetygary-marcus--x

Optimal Posterior Sampling for Policy Identification in Tabular Markov Decision Processes

arXiv:2605.03921v1 Announce Type: new Abstract: We study the (arepsilon, elta)-PAC policy identification problem in finite-horizon episodic Markov Decision Processes. Existing approaches provide finit

Orientation-Aware Unsupervised Domain Adaptation for Brain Tumor Classification Across Multi-Modal MRI

arXiv:2605.03490v1 Announce Type: new Abstract: The clinical integration of deep learning models for brain tumor diagnosis in neuro-oncology is severely constrained by limited expert-annotated MRI dat

Poly-EPO: Training Exploratory Reasoning Models

arXiv:2604.17654v3 Announce Type: replace Abstract: Exploration is a cornerstone of learning from experience: it enables agents to find solutions to complex problems, generalize to novel ones, and sca

Population-Aware Imitation Learning in Mean-field Games with Common Noise

arXiv:2605.03357v1 Announce Type: new Abstract: Mean Field Games (MFGs) provide a powerful framework for modeling the collective behavior of large populations of interacting agents. In this paper, we

Position: Safety and Fairness in Agentic AI Depend on Interaction Topology, Not on Model Scale or Alignment

arXiv:2605.01147v1 Announce Type: new Abstract: As large language models are increasingly deployed as interacting agents in high-stakes decisions, the AI safety community assumes that safety propertie

Power-Softmax: Towards Secure LLM Inference over Encrypted Data

arXiv:2410.09457v2 Announce Type: replace Abstract: Modern cryptographic methods for implementing privacy-preserving LLMs such as gls{HE} require the LLMs to have a polynomial form. Forming such a rep

Predicting missing values: A good idea?

arXiv:2605.03733v1 Announce Type: cross Abstract: Minimizing the Mean Squared Error (MSE) is a key objective in machine learning and is commonly used for imputing missing values. While this approach p

Privacy Preserving Machine Learning Workflow: from Anonymization to Personalized Differential Privacy Budgets in Federated Learning

arXiv:2605.02372v1 Announce Type: cross Abstract: The growing development of artificial intelligence based solutions, together with privacy legislation, has driven the rise of the so-called privacy pr

Pseudo-differential-enhanced physics-informed neural networks

arXiv:2602.14663v2 Announce Type: replace Abstract: We present pseudo-differential enhanced physics-informed neural networks (PINNs), an extension of gradient enhancement but in Fourier space. Gradien

Reinforcement Learning Trained Observer Control for Bearings-Only Tracking

arXiv:2605.02120v1 Announce Type: new Abstract: This paper develops a deep reinforcement learning based observer control policy for autonomous bearings-only tracking of a moving target. The observer m

Resource-Efficient Reinforcement for Reasoning Large Language Models via Dynamic One-Shot Policy Refinement

arXiv:2602.00815v2 Announce Type: replace Abstract: Large language models (LLMs) have exhibited remarkable performance on complex reasoning tasks, with reinforcement learning under verifiable rewards

Rethinking the Rank Threshold for LoRA Fine-Tuning

arXiv:2605.03724v1 Announce Type: new Abstract: A recent landscape analysis of LoRA fine-tuning in the neural tangent kernel regime establishes a sufficient condition r(r+1)/2 > KN on the LoRA rank r

RLDX-1 Technical Report

arXiv:2605.03269v1 Announce Type: cross Abstract: While Vision-Language-Action models (VLAs) have shown remarkable progress toward human-like generalist robotic policies through the versatile intellig

Safety-critical Control Under Partial Observability: Reach-Avoid POMDP meets Belief Space Control

arXiv:2603.10572v2 Announce Type: replace Abstract: Partially Observable Markov Decision Processes (POMDPs) provide a principled framework for robot decision-making under uncertainty. Solving reach-av

Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses

arXiv:2605.02900v1 Announce Type: cross Abstract: Embodied Artificial Intelligence (Embodied AI) integrates perception, cognition, planning, and interaction into agents that operate in open-world, saf

Sample-Efficient Optimization over Generative Priors via Coarse Learnability

arXiv:2503.06917v5 Announce Type: replace Abstract: We study zeroth-order optimization where solutions must minimize a cost d(s) while maintaining high probability under a complex generative prior L(s

SCION: Size-aware Policy Orchestration for Nonstationary Object Caches (Long Paper Version)

arXiv:2605.01055v1 Announce Type: cross Abstract: Object caches underpin cloud and edge services, but production workloads are heterogeneous, nonstationary, and throughput-constrained. Recent simple n

SERE: Structural Example Retrieval for Enhancing LLMs in Event Causality Identification

arXiv:2605.03701v1 Announce Type: new Abstract: Event Causality Identification (ECI) requires models to determine whether a given pair of events in a context exhibits a causal relationship. While Larg

safetyarxiv-cs-cl

Set-Based Training of Neural Barrier Certificates for Safety Verification of Dynamical Systems

arXiv:2605.02526v1 Announce Type: cross Abstract: Barrier certificates are scalar functions over the state space of dynamical systems that separate all unsafe states from all reachable states. The exi

SigLoMa: Learning Open-World Quadrupedal Loco-Manipulation from Ego-Centric Vision

arXiv:2605.03846v1 Announce Type: new Abstract: Designing an open-world quadrupedal loco-manipulation system is highly challenging. Traditional reinforcement learning frameworks utilizing exteroceptio

SMoE: An Algorithm-System Co-Design for Pushing MoE to the Edge via Expert Substitution

arXiv:2508.18983v3 Announce Type: replace Abstract: The Mixture of Experts (MoE) architecture has emerged as a key technique for scaling Large Language Models by activating only a subset of experts pe

SoDa2: Single-Stage Open-Set Domain Adaptation via Decoupled Alignment for Cross-Scene Hyperspectral Image Classification

arXiv:2605.03371v1 Announce Type: new Abstract: Cross-scene hyperspectral image (HSI) classification stands as a fundamental research topic in remote sensing, with extensive applications spanning vari

Steerable Adversarial Scenario Generation through Test-Time Preference Alignment

arXiv:2509.20102v2 Announce Type: cross Abstract: Adversarial scenario generation is a cost-effective approach for safety assessment of autonomous driving systems. However, existing methods are often

Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation

arXiv:2605.03849v1 Announce Type: new Abstract: Distillation-based acceleration has become foundational for making autoregressive streaming video diffusion models practical, with distribution matching

Structured Diffusion Bridges: Inductive Bias for Denoising Diffusion Bridges

arXiv:2605.02973v1 Announce Type: new Abstract: Modality translation is inherently under-constrained, as multiple cross-modal mappings may yield the same marginals. Recent work has shown that diffusio

T^2PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning

arXiv:2605.02178v1 Announce Type: new Abstract: Recent progress in multi-turn reinforcement learning (RL) has significantly improved reasoning LLMs' performances on complex interactive tasks. Despite

Talk is Cheap, Communication is Hard: Dynamic Grounding Failures and Repair in Multi-Agent Negotiation

arXiv:2605.01750v1 Announce Type: cross Abstract: Grounding is the collaborative process of establishing mutual belief sufficient for the current communicative purpose. While static grounding maps lan

TeamUp: Semantic Project Matching and Team Formation for Learning at Scale

arXiv:2605.03237v1 Announce Type: cross Abstract: Project-based learning improves student engagement and learning outcomes, yet allocating students to appropriately challenging projects while forming

safetyarxiv-cs-cl

The AI risk repository: A meta-review, database, and taxonomy of risks from artificial intelligence

arXiv:2408.12622v3 Announce Type: replace-cross Abstract: Artificial intelligence (AI) is reshaping society, from video generation to medical diagnosis, coding agents to autonomous vehicles. Yet resea

The Design and Composition of Structural Causal Decision Processes

arXiv:2605.02681v1 Announce Type: cross Abstract: We present two new classes of causal models of decision-making agents. Our approach is motivated by the needs of modeling the economics of computing s

The Garden of Forking Paths: Narrative Arc-Conditioned Gameplay Planning

arXiv:2605.01245v1 Announce Type: cross Abstract: Narrative archetypes (e.g., Hero's Journey, Three-act structure) provide universal story structures that resonate across cultures and media and are im