safety

Lagrangian Perturbation Diffusion Steering: Latent Reinforcement Learning for Generative Policies

arXiv:2606.01151v1 Announce Type: new Abstract: Behavior cloning with high-capacity generative policies achieves strong imitation performance, but is often limited by demonstration coverage and distri

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Large Language Model Guided Incentive Aware Reward Design for Cooperative Multi-Agent Reinforcement Learning

DGX agent

arXiv:2603.24324v4 Announce Type: replace-cross Abstract: Designing effective auxiliary rewards for cooperative multi-agent systems remains challenging, as misaligned incentives can induce suboptimal

safetyarxiv-cs-ai

2 Jun 2026

Latent Reasoning in TRMs is Secretly a Policy Improvement Operator

arXiv:2511.16886v5 Announce Type: replace-cross Abstract: Recently, small models with latent recursion have obtained promising results on complex reasoning tasks. These results are typically explained

Learning To Sample From Diffusion Models Via Inverse Reinforcement Learning

arXiv:2602.08689v2 Announce Type: replace Abstract: Diffusion models generate samples through an iterative denoising process guided by a pretrained neural network. Once the denoiser is fixed, the samp

Learning When Not to Act: Mitigating Tool Abuse in Agentic Reinforcement Learning

arXiv:2606.02132v1 Announce Type: new Abstract: Agentic reinforcement learning can induce tool abuse, where models overuse external tools even for queries solvable by internal reasoning. Existing appr

LEGS: Fine-Tuning Teleop-Free VLAs for Humanoid Loco-manipulation in an Embodied Gaussian Splatting World

arXiv:2606.01458v1 Announce Type: new Abstract: Training vision-language-action (VLA) policies for humanoid loco-manipulation is constrained by the high cost and complexity of collecting human teleope

Leyline: KV Cache Directives for Agentic Inference

arXiv:2606.01065v1 Announce Type: cross Abstract: Modern KV cache management assumes the chatbot workload: prompts arrive once and the cache grows append-only, so prefix caching and forward-only evict

LFA: Layer Feature Attention for Run-Time Introspection of 2D Object Detectors in Automated Driving

arXiv:2606.00372v1 Announce Type: new Abstract: Reliable object detection is critical for automated driving, yet even state-of-the-art detectors inevitably make errors that can compromise safety. Intr

LinguIUTics at PsyDefDetect: Iterative Imbalance-Aware Fine-tuning of Qwen3-8B for Psychological Defense Mechanism Classification

arXiv:2606.00647v1 Announce Type: cross Abstract: Detecting psychological defense mechanisms in conversational text remains a challenging clinical NLP problem. For the PsyDefDetect 2026 shared task (n

LLM as a Meta-Judge: Synthetic Data for NLP Evaluation Metric Validation

arXiv:2603.09403v2 Announce Type: replace Abstract: Validating evaluation metrics for NLG typically relies on expensive and time-consuming human annotations, which predominantly exist only for English

LLM Trainer: Automated Robotic Data Generation via Demonstration Augmentation using LLMs

arXiv:2509.20070v2 Announce Type: replace Abstract: We present LLM Trainer, a fully automated pipeline that leverages the world knowledge of Large Language Models (LLMs) to transform a small number of

Longitudinal Multimodal Sensing of Physical Activity and Well-Being in Older Adults

arXiv:2606.00345v1 Announce Type: new Abstract: Wearable and mobile sensing technologies enable continuous monitoring of human behavior and health in real-world settings. However, predictive modeling

Lookahead Sample Reward Guidance for Test-Time Scaling of Diffusion Models

arXiv:2602.03211v2 Announce Type: replace-cross Abstract: Diffusion models have demonstrated strong generative performance; however, generated samples often fail to fully align with human intent. This

Looped Transformers with Layer Normalization Provably Learn the Power Method

arXiv:2606.00605v1 Announce Type: new Abstract: Transformers have achieved remarkable success across a wide range of applications, and a growing body of work suggests that part of their strength comes

Lost in Delusion: Examining LLM Safety Under User Delusions and Distress

arXiv:2606.00975v1 Announce Type: new Abstract: LLM chatbots increasingly serve as a first source of support for people in psychological distress, including those whose distress is entangled with delu

Low-Pass Flow Matching

arXiv:2606.02177v1 Announce Type: new Abstract: Flow Matching typically relies on white noise sources, a choice often misaligned with the power spectra of natural data, which tend to decay with freque

Make Mechanistic Interpretability Auditable: A Call to Develop Guidelines via Continuous Collaborative Reviewing

arXiv:2606.00033v1 Announce Type: cross Abstract: While mechanistic interpretability (MI) has produced important insights into neural network internals, the field has yet to establish a standardized s

Markerless Augmented Reality Registration for Surgical Guidance: A Multi-Anatomy Clinical Accuracy Study

arXiv:2511.02086v2 Announce Type: replace Abstract: Purpose: In this paper, we develop and clinically evaluate a depth-only, markerless augmented reality (AR) registration pipeline on a head-mounted d

Market-Based Replanning for Safety-Critical UAV Swarms in Search and Rescue Missions

arXiv:2606.01970v1 Announce Type: new Abstract: Reliable autonomous UAV swarms in Search and Rescue (SAR) missions require fault-tolerant coordination capable of sustaining operations despite agent de

MASCOT: Towards Multi-Agent Socio-Collaborative Companion Systems

arXiv:2601.14230v2 Announce Type: replace-cross Abstract: Multi-agent systems (MAS) are emerging as promising socio-collaborative companions for emotional and cognitive support. However, existing syst

Massive Spikes in LLMs are Bias Vectors: Mechanistic Uncovering and Spike-Free Quantization

arXiv:2606.02288v1 Announce Type: new Abstract: Massive activation spikes in Large Language Models (LLMs) severely degrade quantization by stretching dynamic ranges. While prior hypotheses characteriz

Maybe @ylecun can be automated after all, @SchmidhuberAI?

Gary Marcus poses a question to Yann LeCun and Jürgen Schmidhuber about whether automation of AI systems (possibly referring to AI development or reasoning processes) might be feasible, suggesting a d

safetygary-marcus--x

Measurement Geometry and Design for Trustworthy Generative Inverse Problems

arXiv:2606.02309v1 Announce Type: cross Abstract: Generative models are increasingly used as priors for inverse problems, but their ability to produce realistic images creates a basic trust problem: a

Measuring the Symmetry--Data Exchange Rate

arXiv:2606.01090v1 Announce Type: cross Abstract: Equivariance theory predicts that an architectural symmetry prior reduces sample complexity by a factor of |G|; this is widely cited but rarely measur

Mechanistic Diagnostics of Spatial Lexical Bias in Multimodal Large Language Model Spatial Reasoning

arXiv:2606.01914v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) remain unreliable on spatial multiple-choice questions, and their failures are often attributed to poorly atten

MESA: Improving MoE Safety Alignment via Decentralized Expertise

arXiv:2606.00651v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) architectures scale Large Language Models (LLMs) efficiently, enabling greater capacity with reduced computational cost by dy

Meta-Black-Box Optimization with Ensemble Surrogate Modeling for Robustness-Accuracy Trade-off within SAEA

arXiv:2606.00862v1 Announce Type: cross Abstract: Surrogate-assisted evolutionary algorithms (SAEAs) have been widely used for expensive black-box optimization problems. However, their reliance on rig

Meta expands Teen Accounts safety features to limit harmful content on Instagram, Facebook, and Messenger, including on nutrition, weight lifting, and anxiety (Eli Tan/New York Times)

Eli Tan / New York Times: Meta expands Teen Accounts safety features to limit harmful content on Instagram, Facebook, and Messenger, including on nutrition, weight lifting, and anxiety — The changes,

safetytechmeme

Microsoft releases ASSERT, an open-source framework that lets developers generate and run AI behavior tests using natural-language descriptions (Ram Iyer/TechCrunch)

Ram Iyer / TechCrunch: Microsoft releases ASSERT, an open-source framework that lets developers generate and run AI behavior tests using natural-language descriptions — AI researchers and labs have ad

safetytechmeme

MidSteer: Optimal Affine Framework for Steering Generative Models

arXiv:2605.05220v2 Announce Type: replace-cross Abstract: Steering intermediate representations has emerged as a powerful strategy for controlling generative models, particularly in post-deployment al

Minimax-Optimal Policy Regret in Partially Observable Markov Games

arXiv:2606.02363v1 Announce Type: new Abstract: We study sequential decision-making in partially observable environments against strategic, adaptive opponents, modeled as partially observable Markov g

Mitigating Bias in Locally Constrained Decoding via Tractable Proposals

arXiv:2606.01926v1 Announce Type: new Abstract: Generations from large language models often fail to conform to desired constraints such as JSON schema. Existing locally constrained decoding (LCD) app

Mitigating Perceptual Judgment Bias in Multimodal LLM-as-a-Judge via Perceptual Perturbation and Reward Modeling

arXiv:2606.02578v1 Announce Type: cross Abstract: Recent multimodal large language models have demonstrated strong reasoning ability, yet their reliability as automated evaluators remains limited by a

MobEvolve: An Agentic Self-Evolving Heuristic System for Interpretable Human Mobility Generation

arXiv:2606.01640v1 Announce Type: new Abstract: Human mobility generation aims to synthesize realistic trip chains for target populations based on individual features. Existing paradigms, including de

Model Multiplicity and Predictive Arbitrariness in Recidivism Risk Assessment

arXiv:2606.02198v1 Announce Type: new Abstract: Prediction tasks over individual futures, which are inherently noisy, often admit multiple similarly accurate models. When these models produce differen

MoEIoU: Rethinking Bounding-Box Regression as a Mixture of Experts

arXiv:2606.00844v1 Announce Type: cross Abstract: Bounding-box regression is a fundamental component of object detection, playing a critical role in precise object localization. Existing Intersection-

Morningstar: Get real, SpaceX just isn’t worth a trillion dollars, let alone two.

Gary Marcus argues that SpaceX's valuation is significantly inflated, contending that the company is not worth the trillion-dollar valuations that have been suggested. The critique appears to challeng

safetygary-marcus--x

MOSAIC: Modular Orchestration for Structured Agentic Intelligence and Composition

arXiv:2606.00708v1 Announce Type: new Abstract: Automated data science is a structured model-selection problem. A solution must choose data transformations, feature representations, architecture, trai

Multi-modal Video Representation Alignment for Robust Self-supervised Driver Distraction Detection

arXiv:2606.02352v1 Announce Type: new Abstract: Robust self-supervised learning of multi-modal video representations is critical for real-world applications such as driver distraction detection, where

Multi-Objective Reference-Aligned Machine Unlearning

arXiv:2606.00399v1 Announce Type: new Abstract: Machine unlearning aims to remove the influence of specific training samples while preserving the model's utility. Existing single-objective approaches,

Multi-Objective Reinforcement Learning for Tactical Decision Making for Trucks in Highway Traffic

arXiv:2601.18783v2 Announce Type: replace-cross Abstract: Balancing safety, efficiency, and operational costs in highway driving poses a challenging decision-making problem for heavy-duty vehicles. A

MURMUR: An Efficient Inference System for Long-Form ASR

arXiv:2606.01483v1 Announce Type: cross Abstract: Long-form automatic speech recognition (ASR) requires both high accuracy and low latency, but existing systems force a trade-off between the two. Chun

MViewRouter: Internalizing Geometric Equivariance via Multi-view Alternating Attention for Combinatorial Routing

arXiv:2606.01084v1 Announce Type: cross Abstract: Combinatorial routing problems such as the Traveling Salesman Problem (TSP) and the Capacitated Vehicle Routing Problem (CVRP) are fundamental NP-hard

MyoSem: Aligning Electromyography to Natural-Language Action Semantics for Hand Action Understanding

arXiv:2606.00174v1 Announce Type: cross Abstract: Electromyography (EMG) directly reflects muscle activation and is a key sensing modality for gesture recognition, prosthetic control, and wearable int

NDPP-Grasp: Non-Differentiable Physical Plausibility Constraint-Guided Task-Oriented Dexterous Grasp Generation

arXiv:2606.02432v1 Announce Type: new Abstract: Task-oriented dexterous grasp generation aims to produce dexterous grasp poses that are both physically plausible and functionally suitable for specifie

Network Distributed Multi-Agent Reinforcement Learning for Consensus Control of Quadcopters

arXiv:2606.02107v1 Announce Type: cross Abstract: This paper proposes a Network Distributed Multi-Agent Reinforcement Learning (ND-MARL) framework for quadcopter consensus control. Compared to convent

NEW: Ahead of the SpaceX IPO, we tracked hundreds of promises that Musk has made over the years (FSD, Mars etc). His success rate is…not goo…

NEW: Ahead of the SpaceX IPO, we tracked hundreds of promises that Musk has made over the years (FSD, Mars etc). His success rate is…not good. And it’s getting worse. Some actual data in our months lo

safetygary-marcus--x

Non-Uniform Noise-to-Signal Ratio in the REINFORCE Policy-Gradient Estimator

arXiv:2602.01460v3 Announce Type: replace-cross Abstract: Policy-gradient methods are widely used in reinforcement learning, yet training often becomes unstable or slows down as learning progresses. W