safety

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

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

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

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

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

Towards Safer Large Reasoning Models by Promoting Safety Decision-Making before Chain-of-Thought Generation

arXiv:2603.17368v2 Announce Type: replace Abstract: Large reasoning models (LRMs) achieved remarkable performance via chain-of-thought (CoT), but recent studies showed that such enhanced reasoning cap

TRACE: A Metrologically-Grounded Engineering Framework for Trustworthy Agentic AI Systems in Operationally Critical Domains

arXiv:2605.03838v1 Announce Type: new Abstract: We introduce TRACE, a cross-domain engineering framework for trustworthy agentic AI in operationally critical domains. TRACE combines a four-layer refer

Tracing Like a Clinician: Anatomy-Guided Spatial Priors for Cephalometric Landmark Detection

arXiv:2605.03358v1 Announce Type: new Abstract: When orthodontists trace cephalometric radiographs, they follow a structured workflow: identify the soft tissue profile, partition the skull into anatom

truly scary: anybody, no matter how noble, can be destroyed by misused AI.

truly scary: anybody, no matter how noble, can be destroyed by misused AI. Artificial intelligence can create deceptive videos that rapidly damage a politician's public reputation. A new study reveals

safetygary-marcus--x

Trustworthy AI Suffers from Invariance Conflicts and Causality is The Solution

arXiv:2605.02640v1 Announce Type: new Abstract: As artificial intelligence (AI), including machine learning (ML) models and foundation models (FMs), is increasingly deployed in high-stakes domains, en

Uncertainty-Aware Trip Purpose Inference from GPS Trajectories via POI Semantic Zones and Pareto Calibration

arXiv:2605.01257v1 Announce Type: new Abstract: Large-scale GPS trajectory data offer rich observations of human mobility, yet assigning trip purposes to detected stops remains challenging due to the

Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks

arXiv:2502.04419v3 Announce Type: replace-cross Abstract: Generating synthetic datasets via large language models (LLMs) has emerged as a promising approach to improve LLM performance. However, LLMs i

Understanding Self-Supervised Learning via Latent Distribution Matching

arXiv:2605.03517v1 Announce Type: new Abstract: Self-supervised learning (SSL) excels at finding general-purpose latent representations from complex data, yet lacks a unifying theoretical framework th

Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe

arXiv:2605.03677v1 Announce Type: new Abstract: On-policy distillation (OPD) has recently emerged as an effective post-training paradigm for consolidating the capabilities of specialized expert models