safety

Code Mixologist : A Practitioner's Guide to Building Code-Mixed LLMs

arXiv:2602.11181v2 Announce Type: replace Abstract: Code-mixing and code-switching (CSW) remain challenging phenomena for large language models (LLMs). Despite recent advances in multilingual modeling

CoLA-Flow Policy: Temporally Coherent Imitation Learning via Continuous Latent Action Flow Matching for Robotic Manipulation

DGX agent

arXiv:2601.23087v3 Announce Type: replace Abstract: Learning long-horizon robotic manipulation requires jointly achieving expressive behavior modeling, real-time inference, and stable execution, which

safetyarxiv-cs-ro

12 May 2026

Composing Policy Gradients and Prompt Optimization for Language Model Programs

arXiv:2508.04660v2 Announce Type: replace Abstract: Group Relative Policy Optimization (GRPO) has proven to be an effective tool for post-training language models (LMs). However, AI systems are increa

Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision

arXiv:2509.14234v3 Announce Type: replace Abstract: Where do learning signals come from when there is no ground truth in post-training? We show that inference compute itself can serve as supervision.

Compute Where it Counts: Self Optimizing Language Models

arXiv:2605.10875v1 Announce Type: cross Abstract: Efficient LLM inference research has largely focused on reducing the cost of each decoding step (e.g., using quantization, pruning, or sparse attentio

Conformity Generates Collective Misalignment in AI Agents Societies

arXiv:2605.10721v1 Announce Type: cross Abstract: Artificial intelligence safety research focuses on aligning individual language models with human values, yet deployed AI systems increasingly operate

Consensus Sampling for Safer Generative AI

arXiv:2511.09493v2 Announce Type: replace Abstract: Motivated by undetectable risks in generative AI, we study a general robust aggregation problem: how to aggregate several probability distributions

Constraint-Aware Diffusion Priors for High-Fidelity and Versatile Quadruped Locomotion

arXiv:2605.08804v1 Announce Type: new Abstract: Reinforcement learning combined with imitation learning has significantly advanced biomimetic quadrupedal locomotion. However, scaling these frameworks

Constraint-Aware Reinforcement Learning via Adaptive Action Scaling

arXiv:2510.11491v3 Announce Type: replace-cross Abstract: Safe reinforcement learning (RL) seeks to mitigate unsafe behaviors that arise from exploration during training by reducing constraint violati

Containment Verification: AI Safety Guarantees Independent of Alignment

arXiv:2605.09045v1 Announce Type: new Abstract: Agentic frameworks are the software layer through which AI agents act in the world. Existing safety methods intervene on the model and therefore remain

Continuity Laws for Sequential Models

arXiv:2605.08539v1 Announce Type: cross Abstract: Inductive biases influence the behavior and performance of sequential models. In this work, we study an underexplored inductive bias in sequential mod

Control-Augmented Autoregressive Diffusion for Data Assimilation

arXiv:2510.06637v3 Announce Type: replace-cross Abstract: Despite advances in test-time scaling and diffusion finetuning, guidance for Auto-Regressive Diffusion Models (ARDMs) remains underexplored. W

Core-Halo Decomposition: Decentralizing Large-Scale Fixed-Point Problems

arXiv:2605.08681v1 Announce Type: cross Abstract: We study solving large-scale fixed-point equation (x^star=ar F(x^star)) with decomposition. Standard strict decomposition assigns each agent a disjoin

Cornerstones or Stumbling Blocks? Deciphering the Rock Tokens in On-Policy Distillation

arXiv:2605.09253v1 Announce Type: cross Abstract: While recent work in Reinforcement Learning with Verifiable Rewards (RLVR) has shown that a small subset of critical tokens disproportionately drives

Crosslingual On-Policy Self-Distillation for Multilingual Reasoning

arXiv:2605.09548v1 Announce Type: new Abstract: Large language models (LLMs) have achieved remarkable progress in mathematical reasoning, but this ability is not equally accessible across languages. E

Crowding Out The Noise: Algorithmic Collective Action Under Differential Privacy

arXiv:2505.05707v2 Announce Type: replace Abstract: The integration of AI into daily life has generated considerable attention and excitement, while also raising concerns about automating algorithmic

DAP: Doppler-aware Point Network for Heterogeneous mmWave Action Recognition

arXiv:2605.09604v1 Announce Type: new Abstract: Millimeter-wave (mmWave) radar provides privacy-preserving sensing and is valuable for human action recognition (HAR). Existing mmWave point cloud datas

safetyarxiv-cs-cv

DAPE: Dynamic Non-uniform Alignment and Progressive Detail Enhancement Techniques for Improving the Performance of Efficient Visual Language Models

arXiv:2605.08902v1 Announce Type: cross Abstract: In recent years, pre-trained visual-linguistic models have demonstrated tremendous potential, becoming a crucial foundational framework for numerous d

DARE: Difficulty-Adaptive Reinforcement Learning with Co-Evolved Difficulty Estimation

arXiv:2605.09188v1 Announce Type: cross Abstract: Reinforcement learning improves the reasoning ability of large language models but remains costly and sample-inefficient, as many rollouts provide wea

Data-driven transport modelling without overfit

arXiv:2605.08801v1 Announce Type: new Abstract: Macroscopic transport modelling aims to predict traffic flows after proposed public policy interventions, such as a new road or railway section or a tem

Debugging the Debuggers: Failure-Anchored Structured Recovery for Software Engineering Agents

arXiv:2605.08717v1 Announce Type: cross Abstract: Software engineering agents are increasingly deployed in evaluable engineering environments, yet post-failure recovery remains costly, manual, and ad

Decoupling Endpoint and Semantic Transition Learning for Zero-Shot Composed Image Retrieval

arXiv:2605.08389v1 Announce Type: cross Abstract: Zero-shot composed image retrieval (ZS-CIR) retrieves a target image from a reference image and a text modification without human-annotated CIR triple

Dependency-Aware Discrete Diffusion for Scene Graph Generation

arXiv:2605.09065v1 Announce Type: new Abstract: Scene graphs (SGs) represent objects and their relationships as structured graphs, enabling applications in image generation, robotics, and 3D understan

safetyarxiv-cs-cv

DexWrist: A Robotic Wrist for Constrained and Dynamic Manipulation

arXiv:2507.01008v3 Announce Type: replace Abstract: Development of dexterous manipulation hardware has primarily focused on hands and grippers. However, these end-effectors are often paired with bulky

dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models

arXiv:2605.09291v1 Announce Type: new Abstract: Discrete flow models (DFMs) are a class of flexible generative models for generating discrete data, and diffusion large language models (dLLMs) can be v

DGPO: Beyond Pairwise Preferences with Directional Consistent Groupwise Optimization

arXiv:2605.10863v1 Announce Type: new Abstract: Although Large Language Models (LLMs) have made remarkable progress, current preference optimization methods still struggle to align directional consist

Disentangled Representation Learning via Flow Matching

arXiv:2602.05214v2 Announce Type: replace Abstract: Disentangled representation learning aims to capture the underlying explanatory factors of observed data, enabling a principled understanding of the

Do Linear Probes Generalize Better in Persona Coordinates?

arXiv:2605.09391v1 Announce Type: new Abstract: It is becoming increasingly necessary to have monitors check for harmful behaviors during language model interactions, but text-only monitoring has not

Drum Synthesis from Expressive Drum Grids via Neural Audio Codecs

arXiv:2605.10281v1 Announce Type: cross Abstract: Generating realistic drum audio directly from symbolic representations is a challenging task at the intersection of music perception and machine learn

DuetFair: Coupling Inter- and Intra-Subgroup Robustness for Fair Medical Image Segmentation

arXiv:2605.10521v1 Announce Type: cross Abstract: Medical image segmentation models can perform unevenly across subgroups. Most existing fairness methods focus on improving average subgroup performanc

Dynamic Skill Lifecycle Management for Agentic Reinforcement Learning

arXiv:2605.10923v1 Announce Type: cross Abstract: Large language model agents increasingly rely on external skills to solve complex tasks, where skills act as modular units that extend their capabilit

DynaMiCS: Fine-tuning LLMs with Performance Constraints using Dynamic Mixtures

arXiv:2605.10770v1 Announce Type: new Abstract: Multi-domain fine-tuning of large language models requires improving performance on target domains while preserving performance on constrained domains,

E-TCAV: Formalizing Penultimate Proxies for Efficient Concept Based Interpretability

arXiv:2605.10261v1 Announce Type: new Abstract: TCAV (Testing with Concept Activation Vectors) is an interpretability method that assesses the alignment between the internal representations of a train

Effective Explanations Support Planning Under Uncertainty

arXiv:2605.08406v1 Announce Type: cross Abstract: Explaining how to get from A to B can be challenging. It requires mentally simulating what the listener will do based on what they are told. To captur

EGL-SCA: Structural Credit Assignment for Co-Evolving Instructions and Tools in Graph Reasoning Agents

arXiv:2605.10366v1 Announce Type: new Abstract: Graph reasoning agents operating from natural-language inputs must solve a coupled problem: they must reconstruct a structured graph instance from text,

ElasticFlow: One-Step Physics-Consistent Policy with Elastic Time Horizons for Language-Guided Manipulation

arXiv:2605.08799v1 Announce Type: new Abstract: Diffusion policies have demonstrated exceptional performance in embodied AI. However, their iterative denoising process results in high latency, and exi

Embodied AI in Action: Insights from SAE World Congress 2026 on Safety, Trust, Robotics, and Real-World Deployment

arXiv:2605.10653v1 Announce Type: new Abstract: Embodied artificial intelligence is rapidly moving from research into real-world systems such as autonomous vehicles, mobile robots, and industrial mach

Emergence of Physical Intelligence via Controllable Information Production

arXiv:2601.22449v2 Announce Type: replace Abstract: Intrinsic Motivation (IM) aims to train agents without external rewards, enabling useful behavior to emerge from the agent's interaction with its en

Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning

arXiv:2605.09395v1 Announce Type: new Abstract: In this paper, we propose the first VLnderline{extbf{M}} nderline{extbf{a}}gentic nderline{extbf{r}}easoning framework for few-nderline{extbf{s}}hot mul

Equivariant Reinforcement Learning for Clifford Quantum Circuit Synthesis

arXiv:2605.10910v1 Announce Type: cross Abstract: We consider the problem of synthesizing Clifford quantum circuits for devices with all-to-all qubit connectivity. We approach this task as a reinforce

EROAS: 3D Efficient Reactive Obstacle Avoidance System for Autonomous Underwater Vehicles using 2.5D Forward-Looking Sonar

arXiv:2411.05516v3 Announce Type: replace Abstract: Autonomous Underwater Vehicles (AUVs) have advanced significantly in obstacle detection and path planning through sonar, cameras, and learning-based

ETS: Energy-Guided Test-Time Scaling for Training-Free RL Alignment

arXiv:2601.21484v2 Announce Type: replace Abstract: Reinforcement Learning (RL) post-training alignment for language models is effective, but also costly and unstable in practice, owing to its complic

EvoMAS: Learning Execution-Time Workflows for Multi-Agent Systems

arXiv:2605.08769v1 Announce Type: new Abstract: Large language model (LLM)-based multi-agent systems have shown strong potential on complex tasks through agent specialization, tool use, and collaborat

EvoPref: Multi-Objective Evolutionary Optimization Discovers Diverse LLM Alignments Beyond Gradient Descent

arXiv:2605.09777v1 Announce Type: cross Abstract: Gradient-based preference optimization methods for large language model (LLM) alignment suffer from preference collapse, converging to narrow behavior

EvoStreaming: Your Offline Video Model Is a Natively Streaming Assistant

arXiv:2605.10343v1 Announce Type: cross Abstract: Streaming video understanding demands more than watching longer videos: assistants must decide when to speak in real time, balancing responsiveness ag

Execution Envelopes: A Shared Admission Contract for Backend AI Execution Requests

arXiv:2605.08267v1 Announce Type: cross Abstract: Enterprise AI backends increasingly admit heterogeneous execution requests across model deployment, inference, evaluation, data movement, and agentic

Expert Evaluation and the Limits of Human Feedback in Mental Health AI Safety Testing

arXiv:2601.18061v3 Announce Type: replace Abstract: Learning from human feedback~(LHF) assumes that expert judgments, appropriately aggregated, yield valid ground truth for training and evaluating AI

Explanation-Aware Learning for Enhanced Interpretability in Biomedical Imaging

arXiv:2605.10054v1 Announce Type: new Abstract: Deep neural networks for medical image diagnosis often achieve high predictive accuracy while relying on spurious or clinically irrelevant visual cues,

safetyarxiv-cs-cv