safety

Test-Time Verification for Text-to-SQL via Outcome Reward Models

arXiv:2606.30851v1 Announce Type: cross Abstract: Improving the reliability of large language models (LLMs) at inference time is a central challenge in structured reasoning tasks such as Text-to-SQL.

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

That day when @elonmusk starts to sound a bit like my 2023 TED Talk, which called for “a global, nonprofit organization to regulate the tech…

DGX agent

That day when @elonmusk starts to sound a bit like my 2023 TED Talk, which called for “a global, nonprofit organization to regulate the tech for the sake of democracy and our collective future.” Good

safetygary-marcus--x

1 Jul 2026

The Past Is Prologue: A Plug-in Controller for Selective Updates in Sequentially Evolving LLM Memory

arXiv:2606.31121v1 Announce Type: new Abstract: Sequentially evolving LLM memory enables agents to reuse past experience, but existing systems usually deploy each locally generated memory update witho

Token-Sparse Medical Multimodal Reasoning via Dual-Stream Reinforcement Learning

arXiv:2606.31599v1 Announce Type: cross Abstract: Vision-language models (VLMs) combining reinforcement learning (RL) ignite remarkable progress in multimodal reasoning, yet still struggle with medica

TORA: Topological Representation Alignment for 3D Shape Assembly

arXiv:2604.04050v2 Announce Type: replace Abstract: Flow-matching methods for 3D shape assembly learn point-wise velocity fields that transport parts toward assembled configurations, yet they receive

Toxicity Assessment in Preclinical Histopathology via Class-Aware Mahalanobis Distance for Known and Novel Anomalies

arXiv:2602.02124v2 Announce Type: replace-cross Abstract: Drug-induced toxicity is a leading cause of preclinical and early-clinical failure, making early detection critical. Histopathology is the gol

TraCeS: Learning Per-Timestep Constraint-Violation Credit from Sparse Trajectory-Level Labels

arXiv:2504.12557v3 Announce Type: replace-cross Abstract: Ensuring safe behavior in reinforcement learning (RL) is challenging when safety constraints are implicit and cannot be densely measured. In m

Training Therapeutic Judges and Multi-Agent Systems for Human-Aligned Mental Health Support

arXiv:2606.30887v1 Announce Type: cross Abstract: Large language models show promise for mental health support, yet therapeutic quality improves only when evaluation functions as an actionable control

Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation

arXiv:2606.31184v1 Announce Type: cross Abstract: Adaptive experiments for average treatment effects (ATE) require randomized allocations balancing valid inference with statistical efficiency. The ora

TreeAgent: A Generalizable Multi-Agent Framework for Automated Bias Labeling in Forestry via Compiled Expert Rules and Vision-Language Models

arXiv:2606.31976v1 Announce Type: new Abstract: Human-labeled data are widely used as reference annotations in ML, despite known variability across annotators in many expert-driven domains. In additio

TRIAGE: Role-Typed Credit Assignment for Agentic Reinforcement Learning

arXiv:2606.32017v1 Announce Type: cross Abstract: Agentic reinforcement learning requires assigning credit to environment-facing actions such as searches, clicks, edits, navigation commands, and objec

UniTac: A Unified Multimodal Model for Cross-Sensor Tactile Understanding and Generation

arXiv:2606.31451v1 Announce Type: cross Abstract: Unified multimodal models (UMMs) have shown great promise in integrating understanding and generation across diverse modalities. However, existing res

Unsupervised Data-Efficient Cross-Modal Retrieval with Global-Neighborhood Alignment Hashing

arXiv:2606.31517v1 Announce Type: cross Abstract: Compared to supervised cross-modal hashing (CMH), unsupervised CMH reduces the reliance on manual labeling by learning binary codes from unlabeled ima

Venice AI, which offers access to 200+ AI models while allowing users to retain their privacy, raised a 65M Series A led by Dragonfly at a 1B valuation (Ram Iyer/TechCrunch)

Ram Iyer / TechCrunch: Venice AI, which offers access to 200+ AI models while allowing users to retain their privacy, raised a 65M Series A led by Dragonfly at a 1B valuation — Concerns over the impac

safetytechmeme

Verification-Gated Agentic Mission-State Governance for Intelligent Industrial Multi-Robot Systems

arXiv:2606.31339v1 Announce Type: new Abstract: Agentic artificial intelligence is increasingly used to decompose industrial tasks, propose robot actions, and adapt execution plans in dynamic cyber-ph

VIGOR: VIdeo Geometry-Oriented Reward for Temporal Generative Alignment

arXiv:2603.16271v3 Announce Type: replace Abstract: Video diffusion models lack explicit geometric supervision during training, leading to inconsistency artifacts such as object deformation, spatial d

Vision-Language Procedural Reasoning for Context-Aware Reward Modeling of Robotic Endovascular Guidewire Navigation

arXiv:2606.30698v1 Announce Type: new Abstract: Robotic-assisted endovascular interventions demand accurate, stable, and context-aware guidewire navigation in complex and patient-specific vascular ana

Wait, am I Being Fair? Characterizing Deductive Stereotyping and Mitigating It with Fair-GCG

arXiv:2606.30989v1 Announce Type: cross Abstract: Warning: This paper contains several toxic and offensive statements. While reasoning generally improves fairness in recent large language models (LLMs

Warp RL: Reshaping Base Policy Distributions for Dynamics Adaptation

arXiv:2606.31043v1 Announce Type: new Abstract: Residual reinforcement learning adapts a pretrained robot policy by learning an additive correction to its actions. While effective when adaptation amou

What Probing Reveals about Autonomous Driving: Linking Internal Prediction Errors to Ego Planning

arXiv:2606.31106v1 Announce Type: cross Abstract: Large-scale datasets and fast simulators have enabled improvements in driving policies that appear safe and robust, yet strong performance in nominal

Will the AI bubble pop is the wrong question. We partnered with Damon Cassidy, a video essayist with ~300k subscribers: even if it pops, the…

Will the AI bubble pop is the wrong question. We partnered with Damon Cassidy, a video essayist with ~300k subscribers: even if it pops, the race to uncontrollable superintelligence doesn't go away. P

safetyconnor-leahy--x

A causal modeling perspective on decision theory

arXiv:2606.29911v1 Announce Type: new Abstract: Decision theory provides a formal framework for how agents should make choices under uncertainty, drawing on ideas from philosophy, probability, and cau

A Gravitational Interpretation of Fine-Tuning Reversion

arXiv:2606.28525v1 Announce Type: cross Abstract: Fine-tuning on harmless data can partially undo behaviors acquired earlier in training. Safety can erode under benign post-alignment updates, unlearne

A Hybrid Framework for Song Lyric Annotation Based on Human-LLM Alignment

arXiv:2606.29273v1 Announce Type: cross Abstract: Emotion recognition of song lyrics is a challenging task since lyrics may not necessarily align with the overall emotion of a song. As a result, lyric

A3M: Adaptive, Adversarial and Multi-Objective Learning for Strategic Bidding in Repeated Auctions

arXiv:2606.28943v1 Announce Type: new Abstract: Learning to bid in repeated multi-unit auctions with bandit feedback poses a fundamental challenge. Existing methods often rely on rigid explore-then-ex

AB-RAG: Adaptive Budgeted Retrieval-Augmented Generation for Reliable Question Answering

arXiv:2606.29090v1 Announce Type: cross Abstract: Retrieval-Augmented Generation (RAG) has become the standard way to ground large language models in external knowledge, yet most systems retrieve a fi

AccelAes: Accelerating Diffusion Transformers for Training-Free Aesthetic-Enhanced Image Generation

arXiv:2603.12575v2 Announce Type: replace Abstract: Diffusion Transformers (DiTs) are a dominant backbone for high-fidelity text-to-image generation due to strong scalability and alignment at high res

Accurate Recognition of Pneumonia and COVID-19 by Geometric Shape Normalization of Lung Region using Automatic Landmark Detection and Piecewise Affine Warping

arXiv:2606.29715v1 Announce Type: new Abstract: This paper presents an automatic system for recognizing pulmonary diseases in chest X-rays using geometric normalization of the lung region. The method

ACPO: Agent-Chained Policy Optimization for Multi-Agent Reinforcement Learning

arXiv:2606.30072v1 Announce Type: new Abstract: Cooperative tasks in Multi-Agent Reinforcement Learning (MARL) require agents to collectively maximize a shared return. Under the Centralized Training w

ActiveVital: Geometry-Aware Embodied Vital Signs Monitoring for Home Healthcare Robots

arXiv:2606.30275v1 Announce Type: new Abstract: Home robots require reliable vital signs monitoring to support long-term companionship and safety in daily environments, yet obtaining respiration and h

Adaptive Block Diffusion: Resolving Training-Inference Mismatch in Diffusion Language Models

arXiv:2606.29275v1 Announce Type: new Abstract: Diffusion Language Models (DLMs) are typically trained under fixed context structures, restricting denoising to predetermined token subsets. This create

AERMANI-VLM: Structured Prompting and Reasoning for Aerial Manipulation with Vision Language Models

arXiv:2511.01472v2 Announce Type: replace Abstract: The rapid progress of vision--language models (VLMs) has sparked growing interest in robotic control, where natural language can express the operati

AeroPlace-Flow: Language-Grounded Object Placement for Aerial Manipulators via Visual Foresight and Object Flow

arXiv:2603.07744v2 Announce Type: replace Abstract: Precise object placement remains underexplored in aerial manipulation, where most systems rely on predefined target coordinates and focus primarily

Agent Safety Is Action Alignment

arXiv:2606.28739v1 Announce Type: new Abstract: Large language models increasingly act as agents: they call tools, move money, delete records, and send messages on a user's behalf. To keep them safe,

Agentic Safety is an Epistemic Property, Not a Behavioral One

arXiv:2606.28347v1 Announce Type: cross Abstract: Contemporary AI safety spans pre-training interventions, post-training alignment, deployment-time controls, monitoring, and red-teaming. These methods

Agentic Tool Use in Large Language Models

arXiv:2604.00835v2 Announce Type: replace Abstract: Large language models are increasingly being deployed as autonomous agents yet their real world effectiveness depends on reliable tools for informat

'AI Watermarking': Bridging Policy Discourse and Technical Capabilities

arXiv:2606.28331v1 Announce Type: cross Abstract: The widespread deployment of generative artificial intelligence (AI) models has raised serious concerns about the proliferation of AI-generated conten

An Integrated Machine Learning and Hierarchical Variance Decomposition Pipeline for Student Performance Prediction and Metacognitive Calibration on Multi-Signal Telemetry

arXiv:2606.28881v1 Announce Type: cross Abstract: Predicting student performance and characterizing metacognitive calibration are essential for personalization in intelligent tutoring systems. Prior r

Analytic Concept-Centric Memory for Agentic Embodied Manipulation

arXiv:2606.29774v1 Announce Type: new Abstract: Long-horizon embodied manipulation requires agents to remember persistent objects, track changing scene states, and reuse prior interaction knowledge. H

Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems

arXiv:2606.20470v2 Announce Type: replace-cross Abstract: Agentic AI systems increasingly rely on language-model components to interpret instructions, process external data, invoke tools, and coordina

Anomaly detection using dynamic thresholds and two-year-long alerts in Cloud Monitoring

Choosing the threshold of an alert policy can be a headache. You have to analyze historical data, aggregate it into semantically meaningful time series, and choose a threshold that matters. If the wor

safetygoogle-cloud-ai

AnyBody: Free-Form Whole-Body Humanoid Control from Arbitrary Keypoint Guidance

arXiv:2606.29209v1 Announce Type: cross Abstract: We present AnyBody, a unified whole-body humanoid controller driven by an arbitrary subset of body keypoints chosen at deploy time. Prior physics-base

Are Humans Evolved Instruction Followers? An Underlying Inductive Bias Enables Rapid Instructed Task Learning

arXiv:2606.29792v1 Announce Type: new Abstract: Human adults can often perform a novel task correctly on the first attempt after only receiving verbal or written instructions. This rapid instructed ta

Are We Measuring Strategy or Phrasing? The Gap Between Surface- and Approach-Level Diversity in LLM Math Reasoning

arXiv:2606.29985v1 Announce Type: new Abstract: Diversity in LLM mathematical reasoning is critical for exploration, but common diversity metrics mostly capture surface-level variation rather than dif

Are Whitepaper Claims Reflected in Market Structure? A Contamination-Aware Pipeline and a Power-Limited Null

arXiv:2601.20336v5 Announce Type: replace-cross Abstract: Do the functional narratives in cryptocurrency whitepapers correspond to how their tokens behave in markets? We develop a content-verified, co

ARKD: Adaptive Reinforcement Learning-Guided Bidirectional KL Divergence Distillation for Text Generation

arXiv:2606.29869v1 Announce Type: cross Abstract: Knowledge distillation (KD) is a key technique for compressing Large Language Models (LLMs), yet methods relying on a single KL objective often fail t

Bandwidth Selection in Kernel Density Estimation for Model Calibration

arXiv:2606.29925v1 Announce Type: new Abstract: As deep learning models are increasingly deployed in high-stakes applications, providing well-calibrated uncertainty estimates has become as critical as

Be Faithful When Response: Returning Fluent and Grounded Answers for Vision-Language Models Reinforcement Learning

arXiv:2606.29984v1 Announce Type: new Abstract: Reinforcement Learning (RL) is an important paradigm for improving the reasoning capabilities of Vision-Language Models (VLMs). However, directly applyi