safety

Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective

arXiv:2610.03185v1 Announce Type: new Abstract: On-policy distillation (OPD) has become an important approach to language model post-training. However, despite its performance gains, OPD can also coll

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

GeoScaffold: Learning Compact Geometric Latents via Reconstruction for Efficient Vision-Language Navigation

DGX agent

arXiv:2610.02697v1 Announce Type: cross Abstract: Recent vision-and-language navigation (VLN) systems increasingly adopt streaming Video-LLM policies that map egocentric RGB observations and instructi

safetyarxiv-cs-cv

5 Oct 2026

GTDD: Generative Test-Driven Development for AI Coding Agents with Adversarial Testing

arXiv:2610.02952v1 Announce Type: cross Abstract: Test-driven development gives AI coding agents executable requirements for implementing software. Because these agents can adapt their implementations

hacktrace: behavior-supervised detection of reward hacking during code generation

arXiv:2610.03055v1 Announce Type: new Abstract: A coding agent can earn a passing grade by fixing its code, or by deleting the test that exposes the bug. Detecting such reward hacking requires recogni

Harness-Aware Distillation for Small Language Model Agents

arXiv:2610.02858v1 Announce Type: new Abstract: Language model agents are deployed with a harness, the software around the model that manages its context, tools, and feedback. When such an agent is di

HASTE: Evolving Agent Harnesses Against Emerging Attacks Using Sparse Evidence

arXiv:2610.02920v1 Announce Type: new Abstract: Agent harnesses play a critical role in defenses by enforcing safety constraints to prevent unsafe actions. However, rapidly emerging attacks outpace ma

Heads, Tails, and AI Fails: LLMs, Randomness, and Human Judgments

arXiv:2406.00092v2 Announce Type: replace Abstract: Randomness is central to human cognition and to many applications in which large language models are deployed, yet probabilistic token generation do

How Causality Bridges the Semantic Gap

arXiv:2610.02594v1 Announce Type: cross Abstract: Numerical measurements capture how a system behaves, but often leave the meanings of its variables unspecified. Some variables are measured but never

Hybrid Machine Learning-Assisted Raman Spectroscopy with Generative Feature Augmentation for Pharmaceutical Identification

arXiv:2610.02224v1 Announce Type: new Abstract: Rapid and reliable identification of pharmaceutical residues is important for safeguarding public health, ensuring food safety, and enabling practical R

'I just assumed that it would translate': examining MT risk awareness among healthcare staff with abbreviations as a use case

arXiv:2610.02496v1 Announce Type: new Abstract: In the UK, public healthcare staff report turning to machine translation (MT) - predominantly Google Translate (GT) - to communicate with patients acros

IDRF: Inverse-Distilled Reward Fine-tuning of Masked Discrete Diffusion Models

arXiv:2610.03641v1 Announce Type: new Abstract: Masked discrete diffusion models offer a promising alternative to autoregressive generation, but iterative sampling can be costly, and intractable seque

Imagine the Future, Internalize the Gist: Efficient VLA Reasoning via Internalized Spatiotemporal Imagination

arXiv:2610.02626v1 Announce Type: new Abstract: Vision-language-action (VLA) models increasingly incorporate intermediate reasoning to improve robotic manipulation, yet existing approaches primarily r

Instance-Dependent Regret for CMDPs with Step-Wise Constraints

arXiv:2610.02520v1 Announce Type: cross Abstract: We study online learning in episodic tabular constrained Markov decision processes with step-wise safety constraints. In such a setting, the constrain

Intent-Hiding Jailbreaks: An Information-Theoretic Framework for Compositional Attacks

arXiv:2610.02302v1 Announce Type: cross Abstract: Recent work has shown that large language models (LLMs) can be vulnerable to jailbreak attacks in which harmful intent is obscured through composition

Interpreting at Write Time: A Policy Ablation for Multi-Goal Agent Memory

arXiv:2610.02897v1 Announce Type: new Abstract: A long-running assistant cannot keep everything it has seen, so it summarises. Summarising is not neutral: what is kept is chosen against some notion of

Iterating Consistency Models: Stability, Error Bounds and Noise Schedules

arXiv:2610.03414v1 Announce Type: cross Abstract: Consistency models (CMs) have become a leading approach for generating high-quality samples in few steps. However, adding steps can improve or degrade

Keep It CALM: Analyzing the Limits of Global Unsafety in Text-to-Image Generation

arXiv:2610.02300v1 Announce Type: new Abstract: Training-free safeguards for text-to-image generation often rely on a reusable safety signal, such as an unsafe direction or global toxic subspace, appl

KungfuAthleteBot: learning high-dynamic humanoid motion from video with unified robust recovery

arXiv:2610.03388v1 Announce Type: new Abstract: Video is an abundant, inexpensive source of human motion data that is rich in extreme athletic behaviors. Making it usable for humanoid robots, however,

Labels Override Definitions in Jev-Style Typed Decision Models

arXiv:2610.02586v1 Announce Type: new Abstract: A typed decision model answers a fixed question about an input by returning a probability for each of several caller-defined options. Each option carrie

LatticeSMC: Where to Spend Inference-Time Compute in Chunked Sequence Generators

arXiv:2610.02774v1 Announce Type: new Abstract: Long-form generators for music, motion and video produce sequences chunk by chunk, with each chunk generated by iterative denoising while rewards are de

Learning from Evolving Errors: Adaptive Iterative Repair for On-Policy Distillation

arXiv:2610.02700v1 Announce Type: cross Abstract: On-policy self-distillation (OPSD) supplies dense token-level feedback on trajectories sampled from the student's own policy, a richer training signal

safetyarxiv-cs-cl

Learning Reflexive Behavior for Contact-Rich Manipulation

arXiv:2610.02811v1 Announce Type: new Abstract: During contact-rich manipulation, interactions between a robot and its environment carry information about local geometry: a surface prevents penetratio

Learning Style, Forgetting Semantics: A Case Study of SFT and RFT on Classification Tasks

arXiv:2610.02437v1 Announce Type: cross Abstract: Why does supervised fine-tuning (SFT) lead to more forgetting than reinforcement fine-tuning (RFT), even when all teacher demonstrations are semantica

Learning to Revise Reasoning with Segment-wise On-Policy Distillation

arXiv:2610.02703v1 Announce Type: new Abstract: On-policy distillation (OPD) improves large language model reasoning by training students on their own rollouts with dense token-wise supervision from t

Learning Transferable Policies from Action-free Time Series Through Dynamical Embeddings

arXiv:2610.03065v1 Announce Type: new Abstract: Learning control from action-free recordings is challenging because intervention effects are unobserved and policies may exploit errors in reconstructed

Learning What to Investigate Next: Meta-Reasoning for Long-Horizon Research Agents

arXiv:2610.02525v1 Announce Type: new Abstract: Long-horizon research agents must decide both how to investigate and what to investigate next as evidence accumulates. This is hard to learn because suc

Lexicographic Multi-Objective On-Policy Distillation

arXiv:2610.02359v1 Announce Type: cross Abstract: Reinforcement learning from verifiable rewards (RLVR) usually optimizes answer correctness, yet useful language-model behavior also requires high-qual

Lightweight, Rubric-Guided Trajectory Evaluation for Production AI Agents

arXiv:2610.03315v1 Announce Type: new Abstract: Trajectory evaluation is essential for improving the reliability of LLM-based agents, but production use makes it expensive to run repeatedly. Modern ag

ManifoldSplat: Language-Guided Semantic Shape Editing of 3D Gaussian Head Avatars

arXiv:2610.03599v1 Announce Type: new Abstract: High-fidelity 3D head avatars have reached near-photorealistic quality. While recent methods enable text-driven manipulation, they struggle to provide f

Measure Less, Know More: Self-Supervised Test-Time Feature Acquisition

arXiv:2610.03454v1 Announce Type: cross Abstract: Recent progress in multimodal, high-dimensional learning has enabled foundation models to process heterogeneous, large-scale data. However, at test ti

Mental-R1: Aligning LLM Reasoning for Mental Health Assessment

arXiv:2606.13176v2 Announce Type: replace Abstract: Mental health problems such as anxiety, depression, and suicide remain urgent global challenges, where timely and accurate assessment is critical fo

MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning

arXiv:2610.02824v1 Announce Type: new Abstract: Rubric-based reinforcement learning extends reward-driven optimization to open-ended tasks by assigning partial credit to individual response requiremen

Metropolis-Hastings Dominates Importance Resampling for Policy Composition

arXiv:2610.03480v1 Announce Type: new Abstract: Post-training a large language model (LLM) often requires exploring trade-offs between multiple rewards, but retraining for each trade-off is expensive.

Mind the Refinement Gap: When Safe High-Level Robot Plans Produce Unsafe Executions

arXiv:2610.02662v1 Announce Type: new Abstract: Language-enabled robot systems increasingly combine semantic-graph planning with temporal-logic safety monitors. We investigate a trace-completeness ass

Mitigating Private Data Leakage in LLMs with Whiteout

arXiv:2610.02418v1 Announce Type: cross Abstract: Modern large language models (LLMs) are trained on massive, largely unfiltered datasets, including content scraped from nearly every accessible websit

MixVLA: Adaptive Mixing of Non-Invariant Information for Generalizable Vision-Language-Action Models

arXiv:2610.02898v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models have achieved remarkable advances in robotic manipulation, yet their zero-shot generalization under out-of-distribut

MobiAgent: Dual-Loop Recursive Policy Self-Improvement for Long-Horizon Mobile Manipulation

arXiv:2610.03476v1 Announce Type: cross Abstract: Long-horizon mobile manipulation presents significant challenges due to compounding execution errors and capacity interference between locomotion and

Morality is Contextual: Learning Interpretable Moral Contexts from Human Data with Probabilistic Clustering and Large Language Models

arXiv:2512.21439v2 Announce Type: replace-cross Abstract: A key question in current AI alignment research is how to make AI algorithms learn moral values. Because human morality is highly context-depe

Morpheus: A Morphology-Aware Neural Tokenizer and Word Embedder for Turkish

arXiv:2606.18717v2 Announce Type: replace-cross Abstract: Turkish is agglutinative: meaning is carried by morphemes, yet the subword tokenizers that drive modern language models split words by corpus

Multi-Fidelity Policy Gradients Stabilize Data-Scarce Reinforcement Learning

arXiv:2610.02505v1 Announce Type: cross Abstract: Policy gradient methods for on-policy reinforcement learning (RL) can become unstable when expensive, scarce target-domain data yield noisy gradient e

My dice are conscious!! Shut down every casino now, for slave-trading!! Either that or read what Congressman Lieu is trying to tell you. [Th…

My dice are conscious!! Shut down every casino now, for slave-trading!! Either that or read what Congressman Lieu is trying to tell you. [There are actually multiple sources of randomness in LLMs, inc

safetygary-marcus--x

Native Action-Prior Learning from Videos for World Action Models

arXiv:2610.03391v1 Announce Type: new Abstract: World action models integrate future visual dynamics with robot action prediction, but their scalability remains limited by the need for action-annotate

NEEDLEWORK: Offline Rewriting of Robot Data with Verified Local Stitches

arXiv:2610.02339v1 Announce Type: cross Abstract: Robot demonstrations may contain useful behavior even when individual episodes are inefficient or unsuccessful. Trajectory stitching offers a way to c

Network-in-the-Loop at Scale: GPU-Batched 5G Simulation for Massively Parallel Robot Learning

arXiv:2610.02370v1 Announce Type: cross Abstract: Massively parallel GPU simulators train multi-robot policies in thousands of environments, and many fleets use private Fifth-Generation (5G) networks,

Not Until the Evidence Says So: Teaching LLM Investigators When to Close a Case

arXiv:2610.03190v1 Announce Type: cross Abstract: Accident, defect and outage investigations end with a decision that ordinary question answering never faces: whether the evidence gathered so far is e

On Representational Alignment among Embodied Agents

arXiv:2610.02985v1 Announce Type: new Abstract: Embodied agents interacting with the same physical process may maintain heterogeneous, asynchronous, and observer-relative representations. Rather than

On the Convergence of Success Conditioning for Policy Optimization

arXiv:2610.03642v1 Announce Type: new Abstract: Success conditioning is a strategy for improving decision-making policies in stochastic environments; it updates a policy by increasing the probability

Online Verification of Language Model Responses Under Cost Constraints

arXiv:2610.02632v1 Announce Type: new Abstract: As large language models are increasingly deployed for multi-step reasoning, verifying the correctness of their outputs has become essential for maintai