safety

Probing the Origins of Reasoning Performance: Representational Quality for Mathematical Problem-Solving in RL vs. SFT Fine-Tuned Models

arXiv:2607.26119v1 Announce Type: cross Abstract: Large reasoning models trained via reinforcement learning (RL) have been increasingly shown to outperform their supervised fine-tuned (SFT) counterpar

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Quoting Bruce Schneier

DGX agent

The writing assignments I give my students are gym tasks, not work tasks. I ask them to write policy memos not because the world needs more policy memos. I assign them because the very act of writing,

safetysimon-willison

30 Jul 2026

R-SLPR: Region-based Small-to-Large Point-cloud Registration with Contrastive Learning

arXiv:2607.26583v1 Announce Type: new Abstract: Point-cloud (PC) registration is fundamental to three-dimensional (3D) perception in robotic systems. However, classic registration algorithms falter wh

Recover, Decode, Reguard: Guard-Agnostic Defense Amplification againstEncoded VLM Jailbreaks

arXiv:2607.26574v1 Announce Type: cross Abstract: Safety classifiers ('guards') are the dominant black-box defense for vision-language models, yet they judge an input's surface form, not its meaning:

ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D

arXiv:2607.19321v2 Announce Type: replace-cross Abstract: As AI agents begin to automate AI R&D, we need ways to assess whether their outputs are safe to deploy, even when the agents themselves may be

Rethinking Clinical Relevance in Chest X-ray Machine Learning: How Evaluation References Define Performance

arXiv:2607.26333v1 Announce Type: cross Abstract: Chest X-ray (CXR) machine learning relies heavily on automated evaluation using reference standards that aim to approximate clinical judgment. However

Retrospective Orthogonal Design: Response-Surface Reconstruction from Observational Data

arXiv:2607.26219v1 Announce Type: cross Abstract: Regression estimates from observational data can depend on specification under multicollinearity, while sequential sums of squares (SS) depend on term

RL^2-VLA: Adaptive RL Latent Compositional Steering with Test-Time Scaling for Vision-Language-Action Models

arXiv:2607.26991v1 Announce Type: new Abstract: Despite the impressive visuomotor capabilities enabled by Vision-Language-Action (VLA) models, their performance often degrades on challenging and out-o

safetyarxiv-cs-ro

RLMM-Flow: A Flow-based Mobile Manipulation Framework with Latent-Space Reinforcement Learning

arXiv:2607.26460v1 Announce Type: new Abstract: Mobile manipulation requires generating whole-body action chunks that jointly satisfy goal reaching, collision avoidance, base kinematic constraints, ma

safetyarxiv-cs-ro

Robostreet Flow: A Lightweight, Ultra-Low-Drag Electric Tractor and Four-Truck Hybrid Convoy Architecture for Minimum-Cost Point-to-Point Freight

arXiv:2607.26250v1 Announce Type: new Abstract: Line-haul trucking costs are dominated by three comparably sized components: energy, driver labor, and equipment. Most efficiency technologies address o

SCALPEL: Semantic Cross-modal Alignment via LLM-Powered Encoder Learning for Medical Vision-Language Representation

arXiv:2607.26885v1 Announce Type: new Abstract: Vision-language pre-training (VLP) serves as a cornerstone for medical multimodal representation learning. However, existing medical VLP frameworks are

SciFigAlign: Scoring Scientific Figures by Fine-tuned Alignment of Visuals with Manuscript Evidence

arXiv:2607.27066v1 Announce Type: new Abstract: Scientific figure assessment in peer review differs fundamentally from general image quality evaluation: a figure must be visually legible, faithfully s

Searching for Robust Augmentations to Improve Out-of-Domain Generalization in Dermoscopic Skin Cancer Classification

arXiv:2607.26765v1 Announce Type: new Abstract: Background/Objectives: Dermoscopic skin lesion classifiers often lose accuracy under domain shift across imaging devices, illumination, and capture arti

Self-Adaptive Learning and Model Predictive Control for Tracking Unknown Dynamics with No Regret

arXiv:2607.26370v1 Announce Type: cross Abstract: We propose a self-adaptive online learning for control method for tracking unknown target dynamics. The target dynamics can exhibit switching behavior

Shape-Based Inductive Bias for Glioma Grading from Tumor Contours

arXiv:2607.26090v1 Announce Type: cross Abstract: Glioma grading from tumor contours is often treated as a pixel problem even when the signal of interest is shape. We align closed contours with a func

Situational Awareness, the $20bn hedge fund founded by former OpenAI employee Leopold Aschenbrenner, has sought to raise fresh capital from …

Situational Awareness, the $20bn hedge fund founded by former OpenAI employee Leopold Aschenbrenner, has sought to raise fresh capital from investors after suffering heavy losses during the recent rou

safetygary-marcus--x

SkillRise: Agentic Reinforcement Learning for Cross-Task Skill Evolution

arXiv:2607.26784v1 Announce Type: new Abstract: Large language model agents often encounter related yet distinct tasks that share reusable solution patterns. Yet standard agentic reinforcement learnin

SMSP: A Plug-and-Play Strategy of Multi-Scale Perception for MLLMs to Perceive Visual Illusions

arXiv:2603.23118v2 Announce Type: replace Abstract: Recent works have shown that multimodal large language models (MLLMs) are highly vulnerable to hidden-pattern visual illusions, where the hidden con

Stable and Budget-Feasible Coalition Formation for Clustered Federated Learning: A Hedonic Potential-Game Approach

arXiv:2607.26788v1 Announce Type: cross Abstract: Clustered federated learning benefits from organizing heterogeneous participants into coalitions that train coalition-specific models, but such cluste

Steering Instruction Hierarchies at Inference Time

arXiv:2607.26228v1 Announce Type: new Abstract: Instruction hierarchies are a core safety assumption of language model deployment: higher priority inputs, such as system prompts, should override confl

SymmGrid: Super-Scaling On-Robot Learning with Parallelized Symmetries and Egocentric-Exocentric Visual Perception

arXiv:2607.26985v1 Announce Type: cross Abstract: Deep reinforcement policy learning directly in physical robots (on-robot learning) remains bottlenecked by slow wall-clock training times. We present

The Advantage of Fine-Grained Training

arXiv:2509.05130v2 Announce Type: replace Abstract: In classification problems, models are trained to predict a class label based on the input data features. However, class labels are organized hierar

// The agent is its own best speculator // Agents spend a large share of wall-clock time waiting on tool results. Speculation hides that lat…

// The agent is its own best speculator // Agents spend a large share of wall-clock time waiting on tool results. Speculation hides that latency by predicting and pre-executing the next call, but exte

safetydair-ai--x

The Confounder Trap: Treatment-Encoding Representations in Causal Inference with Text

arXiv:2607.26309v1 Announce Type: cross Abstract: Estimating causal effects of linguistic properties from observational text is difficult because the same document can contain both the treatment of in

The narrative: Blame the Agent, instead of the Agency that told him to “apply all your powers and told to achieve this win.” Ironically, man…

The narrative: Blame the Agent, instead of the Agency that told him to “apply all your powers and told to achieve this win.” Ironically, many forefront members of the AI-Safety community, in their fer

safetyyann-lecun--x

Thinking Under Uncertainty: Evidence Use and Information-Seeking in Language Models

arXiv:2607.26845v1 Announce Type: new Abstract: Inference-time thinking improves the performance of large language models, but aggregate outcomes do not reveal whether models use available evidence mo

Towards Grounded GI Endoscopy VQA via Multi-Task Learning on Small VLMs

arXiv:2607.27122v1 Announce Type: new Abstract: Gastrointestinal (GI) endoscopic image analysis has shifted from single-label classification toward visual question answering (VQA), where a model must

TraceCLIP: Recovering Local Semantics from Patch-to-CLS Contributions

arXiv:2607.26107v1 Announce Type: new Abstract: Dense vision-language understanding, including object localization, region recognition, and open-vocabulary semantic segmentation, requires associating

Veritas++: Value-aware On-Policy Distillation for Perception-Enhanced AIGI Detection

arXiv:2607.27113v1 Announce Type: new Abstract: The growing capability of image generation models has made synthetic images a routine presence in open media, making robust and generalizable AI-Generat

Visibility-Aware Cooperative Tracking with Decentralized LiDAR-Based Aerial Swarms

arXiv:2512.01280v2 Announce Type: replace Abstract: Autonomous aerial tracking with drones offers vast potential for surveillance, cinematography, and industrial inspection applications. While single-

safetyarxiv-cs-ro

Weak-to-Strong On-Policy Distillation

arXiv:2607.26246v1 Announce Type: new Abstract: On-policy distillation (OPD), which aligns a student with the teacher's token-level distribution on the student's own rollouts, is an effective paradigm

WikiLoop: Jointly Learning to Build and Navigate Agent-Native Wikis with Downstream Feedback

arXiv:2607.26604v1 Announce Type: new Abstract: Knowledge-base construction and querying are typically optimized in isolation: retrieval-augmented agents operate over a fixed, externally maintained in

A context-adaptive policy framework for robust and reactive robotic manipulation via uncertainty-aware imitation learning

arXiv:2410.24035v2 Announce Type: replace-cross Abstract: Generating robust and reactive manipulation strategies that can adapt to changing context information is a challenging task in robotics. Over

A Functional Approach to Curve Alignment and Shape Analysis

arXiv:2503.05632v2 Announce Type: cross Abstract: In many image analysis problems, the contours of objects carry important statistical information about shape. Such contours are typically affected by

A Unified Algorithmic Framework for Hybrid Reinforcement Learning in Tabular MDPs with Shifted Transition Dynamics

arXiv:2607.25207v1 Announce Type: new Abstract: This paper investigates a hybrid reinforcement learning setting in tabular Markov Decision Processes (MDPs), where an agent aims to learn an optimal pol

A2D2: Audi Autonomous Driving Dataset

arXiv:2004.06320v2 Announce Type: replace Abstract: Research in machine learning, mobile robotics, and autonomous driving is accelerated by the availability of high quality annotated data. To this end

AdaKP: Online Adaptive Knowledge-Point Selection for Reasoning-Oriented Reinforcement Learning

arXiv:2607.24833v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards is a powerful paradigm for eliciting reasoning in large language models, yet it suffers from severe rewar

Agentic AI in medicine: architectures, applications, evaluation, and challenges for clinical translation

arXiv:2607.25489v1 Announce Type: new Abstract: Large language models and multimodal foundation models are enabling medical artificial intelligence (AI) systems to move beyond isolated prediction and

AlphaCrafter: Harnessing Multi-Agent Workflows for Cross-Sectional Quantitative Trading

arXiv:2605.05580v2 Announce Type: replace Abstract: Quantitative trading agents have demonstrated substantial promise in automating factor discovery, signal aggregation, and portfolio execution. Howev

Architectural Backdoors in Vision-Language Model Supply Chains via Representation Steering

arXiv:2607.25479v1 Announce Type: cross Abstract: Vision--Language Models (VLMs) are increasingly deployed through a model supply chain in which pretrained checkpoints, architecture definitions, text

Behavior-Driven Explainability

arXiv:2607.24881v1 Announce Type: new Abstract: As system complexity has vastly increased, it has become significantly more challenging for a single person or a team to fully understand all aspects of

Beyond Predictive Accuracy: A Reliability-Aware Audit of Molecular Representations for Human Olfaction

arXiv:2607.24848v1 Announce Type: cross Abstract: Pretrained molecular encoders are commonly evaluated through downstream prediction, but predictive accuracy alone does not establish that a learned re

Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMs

arXiv:2607.25600v1 Announce Type: cross Abstract: Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unn

Breaking the Curse of Repulsion: Remoteness-Aware Control of Negative Off-Policy Updates

arXiv:2602.10430v2 Announce Type: replace-cross Abstract: Off-policy policy optimization reuses historical behavior, including negative-advantage samples that suppress known failures. We show that rep

Calibrated Partial Resets: Preventing Policy Collapse in Continual Reinforcement Learning

arXiv:2607.24996v1 Announce Type: cross Abstract: Neural networks are hindered by accumulating dormant neurons and loss of expressivity throughout training, particularly in non-stationary data setting

CAST: Game Solvers as Turn-Level Teachers for LLM Agents

arXiv:2607.25308v1 Announce Type: cross Abstract: Training large language models (LLMs) to act in long-horizon games is a promising step toward generalist decision-making, yet reinforcement learning w

CIFNet: An Analytic Neural Learning Framework for Efficient and Calibrated Class-Incremental Learning

arXiv:2509.11285v2 Announce Type: replace-cross Abstract: Class-Incremental Learning (CIL) in deep neural networks is conventionally framed as an iterative gradient-based optimization problem, incurri

Construction-Driven Injection: Linguistically-Grounded Edit-Based Code-Mixing Fingerprints for Large Language Models

arXiv:2607.25633v1 Announce Type: cross Abstract: Large language models (LLMs) are costly intellectual assets that remain exposed to unauthorized redistribution and commercial misuse. Injected fingerp