safety

Certified Speculative Execution for Untrusted AI Agents

arXiv:2606.31023v1 Announce Type: cross Abstract: Hard-constrained sequential decision systems have no certified way to spend the test-time compute of modern AI: executing the multi-step drafts of a l

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

ChronoFlow-Policy: Unifying Past-Current-Future Interaction Flow in Visuomotor Policy Learning

DGX agent

arXiv:2606.31493v1 Announce Type: new Abstract: Visual signals play a crucial role in policy learning by enabling models to capture object motion and interaction dynamics. Just as humans reason about

safetyarxiv-cs-ro

1 Jul 2026

CLExEval: A Human-in-the-Loop Framework for Qualitative Evaluation of LLM Clinical Reasoning

arXiv:2606.31608v1 Announce Type: new Abstract: Large Language Models (LLMs) achieve strong results on many medical benchmarks, but their clinical reasoning remains difficult to evaluate reliably. A c

safetyarxiv-cs-cl

CLOUDADV: Decision-Aligned Instance Sizing with Zero-Shot Foundation Models under Drift

arXiv:2606.31470v1 Announce Type: new Abstract: Cloud virtual machines are often overprovisioned, creating avoidable cost and operational inefficiency. We present CLOUDADV, an interactive engineer-fac

Corruption Robust Offline Reinforcement Learning with Human Feedback

arXiv:2402.06734v2 Announce Type: replace-cross Abstract: We study data corruption robustness for reinforcement learning with human feedback (RLHF) in an offline setting. Given an offline dataset of p

Cross-Space Distillation: Teaching One-Step Students with Modern Diffusion Teachers

arXiv:2606.32020v1 Announce Type: new Abstract: Modern one-step diffusion models achieve impressive quality through distribution-based timestep distillation. Yet, they rely on a critical assumption: T

DataEvolver: Self-Evolving Multi-Agent Data Construction for Text-Rich Image Generation

arXiv:2606.31537v1 Announce Type: new Abstract: Text-rich image generation is one of the most challenging settings in image generation, since models must simultaneously produce visually realistic imag

DDIAgents: Mechanism-Conditioned Context Flow for Drug-Drug Interaction Prediction

arXiv:2606.31085v1 Announce Type: new Abstract: Drug-drug interaction (DDI) prediction is essential for medication safety, yet it requires reasoning over heterogeneous biomedical evidence whose releva

Deep Reinforcement Learning for Spacecraft Attitude Control During Atmospheric Re-Entry

arXiv:2606.31291v1 Announce Type: new Abstract: Deep reinforcement learning has the potential to solve attitude control problems more adaptively, precisely, and robustly by handling nonlinear dynamics

Designing Privacy-Preserving Visual Perception for Robot Navigation Based on User Privacy Preferences

arXiv:2604.06382v2 Announce Type: replace Abstract: Visual navigation is a fundamental capability of mobile service robots, yet the onboard cameras required for such navigation can capture privacy-sen

DISCOVER: A Solver for Distributional Counterfactual Explanations

arXiv:2603.16436v2 Announce Type: replace Abstract: Counterfactual explanations (CE) explain model decisions by identifying input modifications that lead to different predictions. Most existing method

ECHO: Prune to act, trace to learn with selective turn memory in agentic RL

arXiv:2606.31650v1 Announce Type: cross Abstract: Long-horizon language agents must repeatedly interact with tools, accumulate evidence, and make decisions under bounded context windows. Existing cont

Efficient Sim-to-Real Transfer of World-Action Models from Synthetic Priors

arXiv:2606.31101v1 Announce Type: new Abstract: Bridging the sim-to-real gap is a core challenge in deploying learned manipulation policies. Sim-to-real learning is attractive because it can replace e

EgoVITA: Learning to Plan and Verify for Egocentric Video Reasoning

arXiv:2511.18242v3 Announce Type: replace Abstract: Egocentric video understanding requires procedural reasoning under partial observability and continuously shifting viewpoints. Current multimodal la

ELASTIC: Efficiently Learning to Adaptively Scale Test-Time Compute for Generative Control Policies

arXiv:2606.31132v1 Announce Type: new Abstract: Generative control policies (GCPs), such as diffusion policies and flow-based vision-language-action models, enable test-time scaling in robot control.

End-to-End Efficient RL for Linear Bellman Complete MDPs with Deterministic Transitions

arXiv:2603.23461v2 Announce Type: replace Abstract: We study reinforcement learning (RL) with linear function approximation in Markov Decision Processes (MDPs) satisfying linear Bellman completeness -

ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs

arXiv:2606.31982v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) incur prohibitive inference costs due to long visual token sequences. Training-free visual token reduction prov

Evil Spectra: How Optimisers can Amplify or Suppress Emergent Misalignment

arXiv:2606.31591v1 Announce Type: cross Abstract: Emergent misalignment (EM) is a recently discovered phenomenon in LLMs where fine-tuning on a narrow misaligned task, such as writing insecure code, l

Evo-PI: Aligning Medical Reasoning via Evolving Principle-Guided Supervision

arXiv:2606.31800v1 Announce Type: new Abstract: Despite recent progress, the reasoning capabilities of large multimodal language models (MLLMs) remain fundamentally constrained by static supervision,

Explainable Artificial Intelligence For The Detection and Characterisation of Stage B Heart Failure

arXiv:2606.30665v1 Announce Type: cross Abstract: Stage B heart failure is characterized by asymptomatic structural or functional cardiac abnormalities. Identifying individuals at this stage is clinic

ExPLoRe: Expert Patch-Level Loss Routing for Multi-Objective Masked Image Modeling

arXiv:2606.31201v1 Announce Type: new Abstract: Multi-objective masked image modeling (MIM) combines complementary learning signals (token distillation, CLS alignment, and pixel reconstruction) but ex

FastDSAC: Enhancing Policy Plasticity via Constrained Exploration for Scalable Humanoid Locomotion

arXiv:2606.31691v1 Announce Type: new Abstract: Scalable reinforcement learning has popularized high-throughput sampling architectures, which significantly compresses the training time for off-policy

Fine-grained Motion Retrieval via Joint-Angle Motion Images and Token-Patch Late Interaction

arXiv:2603.09930v2 Announce Type: replace Abstract: Text-motion retrieval aims to learn a semantically aligned latent space between natural language descriptions and 3D human motion skeleton sequences

Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models

arXiv:2603.12893v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning f

FLARE-AI: Flaw Reporting for AI

arXiv:2606.31567v1 Announce Type: cross Abstract: Flaw reporting for deployed AI systems is fundamental to identifying system failures and improving AI safety. Yet the AI reporting ecosystem is fragme

Freeform Preference Learning for Robotic Manipulation

arXiv:2606.32027v1 Announce Type: cross Abstract: Reward design remains a central bottleneck for autonomous robot policy improvement, especially in long-horizon manipulation tasks where sparse success

From Failure to Alignment: A Requirements Engineering Framework for Machine Learning Systems

arXiv:2606.31589v1 Announce Type: cross Abstract: Organisations designing, developing, and deploying machine learning systems (MLS) need to be able to check that these systems are trustworthy, and com

From Propositional to Perceptual Asymmetry: Extending Frictive Policy Optimization to Asymmetric Partial Information Dialogue

arXiv:2606.30973v1 Announce Type: new Abstract: Frictive Policy Optimization (FPO; Pustejovsky et al., 2025) treats friction in collaborative dialogue -- misalignment, misunderstanding, repair -- as a

safetyarxiv-cs-cl

G2P: Gaussian-to-Point Attribute Alignment for Boundary-Aware 3D Segmentation

arXiv:2601.03510v3 Announce Type: replace Abstract: Point cloud segmentation is critical for 3D scene understanding. However, sparse and irregular point distributions provide limited appearance eviden

GEAR: Guided End-to-End AutoRegression for Image Synthesis

arXiv:2606.32039v1 Announce Type: new Abstract: Visual generative models are typically trained in two stages. A tokenizer is first trained for reconstruction and then frozen, after which a generator i

GR2 Technical Report

arXiv:2606.31984v1 Announce Type: cross Abstract: Industrial recommendation systems serve billions of users through a multi-stage funnel -- retrieval, early-stage ranking, and re-ranking -- where the

GRAPE: Graph-Augmented Prototype Explanations for Interactive Medical Image Diagnosis

arXiv:2606.30901v1 Announce Type: new Abstract: Prototype-based medical image classifiers present three clinical limitations: they treat findings as independent, silently amplify unsafe physician feed

Great new episode with @So8res on @PeterMcCormack's podcast, give it a watch! https://www.youtube.com/watch?v=1laTnwAdaLA

Connor Leahy promoted a podcast episode featuring So8res (a researcher/AI safety figure) on Peter McCormack's podcast, sharing a YouTube link to the full video on X/Twitter. The post encourages viewer

safetyconnor-leahy--x

GUIDE: Resolving Domain Bias in GUI Agents through Real-Time Web Video Retrieval and Plug-and-Play Annotation

arXiv:2603.26266v3 Announce Type: replace Abstract: Large vision-language models have endowed GUI agents with strong general capabilities for interface understanding and interaction. However, due to i

Harnessing Textual Refusal Directions for Multimodal Safety

arXiv:2606.31876v1 Announce Type: new Abstract: To improve safety in Large Language Models (LLMs) we can either perform post-training alignment or exploit refusal directions in the activation space. B

HistoriQA-ThirdRepublic: Multi-Hop Question Answering Corpus for Historical Research, Parliamentary Debates from the French Third Republic (1870-1940)

arXiv:2606.31325v1 Announce Type: new Abstract: We present HistoriQA-ThirdRepublic: a French-language dataset of multi-hop historical questions derived from parliamentary debates and newspapers of the

How Human Feedback Shapes AI-generated Community Notes

arXiv:2606.30905v1 Announce Type: cross Abstract: Community Notes, a bridging-based crowd-sourced fact-checking system, has emerged as a new mechanism for moderating misleading information on social m

Incentive Aware AI Regulations: A Credal Characterisation

arXiv:2603.05175v2 Announce Type: replace Abstract: The rapid proliferation of AI applications has intensified debate on effective regulation of these black-box services. Effective regulation must bal

InfiniVerse: Occupancy Guided Unbounded Scene Generation for Autonomous Driving

arXiv:2606.31109v1 Announce Type: new Abstract: Generating realistic, controllable, and temporally coherent urban environments is a critical yet unresolved challenge in the autonomous driving communit

IterCAD: An Iterative Multimodal Agent for Visually-Grounded CAD Generation and Editing

arXiv:2606.13368v2 Announce Type: replace Abstract: Computer-Aided Design is pivotal in modern manufacturing, yet existing automated methods predominantly rely on open-loop, one-shot generation, creat

Keep Policy Gradient in Charge: Sibling-Guided Credit Distillation for Long-Horizon Tool-Use Agents

arXiv:2606.12634v2 Announce Type: replace-cross Abstract: Long-horizon tool-use reinforcement learning learns from outcome verification, but trajectory-level advantages are broadcast over reasoning, A

LabGuard: Grounding Natural-Language Laboratory Rules into Runtime Guards for Embodied Laboratory Agents

arXiv:2606.31045v1 Announce Type: new Abstract: Scientific embodied agents are increasingly capable of carrying out laboratory procedures, but executing these procedures safely in dynamic laboratory e

LaMP: Learning Vision-Language-Action Policy with 3D Scene Flow as Latent Motion Prior

arXiv:2603.25399v2 Announce Type: replace Abstract: We introduce extbf{LaMP}, a dual-expert Vision-Language-Action framework that embeds dense 3D scene flow as a latent motion prior for robotic manipu

Language-Assisted Super-Resolution from Real-World Low-Resolution Patches

arXiv:2606.31363v1 Announce Type: new Abstract: Single image super-resolution aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs. Training SR models typically requires pai

Layout-Conditioned Autoregressive Text-to-Image Generation via Structured Masking

arXiv:2509.12046v2 Announce Type: replace-cross Abstract: Although autoregressive (AR) models have demonstrated remarkable success in image generation, extending these models to layout-conditioned gen

Learning Dexterous Grasping from Sparse Taxonomy Guidance

arXiv:2604.04138v2 Announce Type: replace-cross Abstract: Dexterous manipulation requires planning a grasp configuration suited to the object and task, which is then executed through coordinated multi

Learning Structurally Consistent Representations for Multi-View Radar Semantic Segmentation

arXiv:2606.31609v1 Announce Type: cross Abstract: Radar sensors provide reliable perception under adverse weather and lighting conditions, but their sparse, noisy, and weakly semantic measurements mak

Learning Where to Look: A Reinforcement Learning Framework for Robust Micro-Ultrasound Prostate Cancer Detection

arXiv:2606.30951v1 Announce Type: cross Abstract: Micro-ultrasound (muUS) is a new, emerging, and promising imaging modality for prostate cancer (PCa) detection, but accurate identification of suspici