safety

Discovering a Shared Logical Subspace: Steering LLM Logical Reasoning via Alignment of Natural-Language and Symbolic Views

arXiv:2604.19716v1 Announce Type: new Abstract: Large Language Models (LLMs) still struggle with multi-step logical reasoning. Existing approaches either purely refine the reasoning chain in natural l

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Do Emotions Influence Moral Judgment in Large Language Models?

DGX agent

arXiv:2604.19125v1 Announce Type: new Abstract: Large language models have been extensively studied for emotion recognition and moral reasoning as distinct capabilities, yet the extent to which emotio

safetyarxiv-cs-cl

22 Apr 2026

Documents and sources: insurers including QBE and Beazley are moving to cap cyber policy payouts for losses and regulatory fines tied to AI use and 'LLMjacking' (Lee Harris/Financial Times)

Lee Harris / Financial Times: Documents and sources: insurers including QBE and Beazley are moving to cap cyber policy payouts for losses and regulatory fines tied to AI use and “LLMjacking” — Beazley

safetytechmeme

DT2IT-MRM: Debiased Preference Construction and Iterative Training for Multimodal Reward Modeling

arXiv:2604.19544v1 Announce Type: new Abstract: Multimodal reward models (MRMs) play a crucial role in aligning Multimodal Large Language Models (MLLMs) with human preferences. Training a good MRM req

Dual Triangle Attention: Effective Bidirectional Attention Without Positional Embeddings

arXiv:2604.18603v1 Announce Type: cross Abstract: Bidirectional transformers are the foundation of many sequence modeling tasks across natural, biological, and chemical language domains, but they are

Efforts to revive chip manufacturing in Pennsylvania have been left in limbo by President Trump's sudden upending of US semiconductor policy over the past year (Michael Acton/Financial Times)

Michael Acton / Financial Times: Efforts to revive chip manufacturing in Pennsylvania have been left in limbo by President Trump's sudden upending of US semiconductor policy over the past year — High-

safetytechmeme

Enhancing Construction Worker Safety in Extreme Heat: A Machine Learning Approach Utilizing Wearable Technology for Predictive Health Analytics

arXiv:2604.19559v1 Announce Type: new Abstract: Construction workers are highly vulnerable to heat stress, yet tools that translate real-time physiological data into actionable safety intelligence rem

Ensembling Pruned Attention Heads For Uncertainty-Aware Efficient Transformers

arXiv:2510.18358v2 Announce Type: replace-cross Abstract: Uncertainty quantification (UQ) is essential for deploying deep neural networks in safety-critical settings. Although methods like Deep Ensemb

Evaluating LLM-Driven Summarisation of Parliamentary Debates with Computational Argumentation

arXiv:2604.19331v1 Announce Type: new Abstract: Understanding how policy is debated and justified in parliament is a fundamental aspect of the democratic process. However, the volume and complexity of

EVPO: Explained Variance Policy Optimization for Adaptive Critic Utilization in LLM Post-Training

arXiv:2604.19485v1 Announce Type: cross Abstract: Reinforcement learning (RL) for LLM post-training faces a fundamental design choice: whether to use a learned critic as a baseline for policy optimiza

ExpertGen: Scalable Sim-to-Real Expert Policy Learning from Imperfect Behavior Priors

arXiv:2603.15956v2 Announce Type: replace-cross Abstract: Learning generalizable and robust behavior cloning policies requires large volumes of high-quality robotics data. While human demonstrations (

Failure Modes in Multi-Hop QA: The Weakest Link Effect and the Recognition Bottleneck

arXiv:2601.12499v2 Announce Type: replace Abstract: Despite scaling to massive context windows, Large Language Models (LLMs) struggle with multi-hop reasoning due to inherent position bias, which caus

Fairness Audits of Institutional Risk Models in Deployed ML Pipelines

arXiv:2604.19468v1 Announce Type: cross Abstract: Fairness audits of institutional risk models are critical for understanding how deployed machine learning pipelines allocate resources. Drawing on mul

FairTree: Subgroup Fairness Auditing of Machine Learning Models with Bias-Variance Decomposition

arXiv:2604.19357v1 Announce Type: new Abstract: The evaluation of machine learning models typically relies mainly on performance metrics based on loss functions, which risk to overlook changes in perf

Fascinating how AI is getting better at diagrams like these (at least for ones that you could easily find on web search) but still making so…

Fascinating how AI is getting better at diagrams like these (at least for ones that you could easily find on web search) but still making some pretty wacky errors — like confusing where the rear brake

FASE : A Fairness-Aware Spatiotemporal Event Graph Framework for Predictive Policing

arXiv:2604.18644v1 Announce Type: cross Abstract: Predictive policing systems that allocate patrol resources based solely on predicted crime risk can unintentionally amplify racial disparities through

FASTER: Value-Guided Sampling for Fast RL

arXiv:2604.19730v1 Announce Type: cross Abstract: Some of the most performant reinforcement learning algorithms today can be prohibitively expensive as they use test-time scaling methods such as sampl

FB-NLL: A Feature-Based Approach to Tackle Noisy Labels in Personalized Federated Learning

arXiv:2604.19729v1 Announce Type: new Abstract: Personalized Federated Learning (PFL) aims to learn multiple task-specific models rather than a single global model across heterogeneous data distributi

Filing: Tron founder Justin Sun sues the Trump family's World Liberty Financial, alleging it unfairly locked up his WLFI holdings and threatened and defamed him (CoinDesk)

CoinDesk: Filing: Tron founder Justin Sun sues the Trump family's World Liberty Financial, alleging it unfairly locked up his WLFI holdings and threatened and defamed him — World Liberty unfairly froz

safetytechmeme

Fitted Q Evaluation Without Bellman Completeness via Stationary Weighting

arXiv:2512.23805v2 Announce Type: replace-cross Abstract: Fitted Q-evaluation (FQE) is a foundational method for off-policy evaluation in reinforcement learning, but existing theory typically relies o

Framelet-Based Blind Image Restoration with Minimax Concave Regularization

arXiv:2604.19314v1 Announce Type: new Abstract: Recovering corrupted images is one of the most challenging problems in image processing. Among various restoration tasks, blind image deblurring has bee

From Particles to Perils: SVGD-Based Hazardous Scenario Generation for Autonomous Driving Systems Testing

arXiv:2604.18918v1 Announce Type: cross Abstract: Simulation-based testing of autonomous driving systems (ADS) must uncover realistic and diverse failures in dense, heterogeneous traffic. However, exi

Gives new meaning to “Rear Brake Lever”!

This post likely references a humorous or unexpected use case involving a rear brake lever, possibly demonstrating an unintended design flaw, unconventional application, or double meaning related to b

God these people are annoying. Obnoxious comment and the guy can’t be bothered to notice the front tire that is labeled as a fork 🙄 Or to n…

God these people are annoying. Obnoxious comment and the guy can’t be bothered to notice the front tire that is labeled as a fork 🙄 Or to notice the front brake that’s lost its cable and is hovering u

GRAIL:Learning to Interact with Large Knowledge Graphs for Retrieval Augmented Reasoning

arXiv:2508.05498v2 Announce Type: replace Abstract: Large Language Models (LLMs) integrated with Retrieval-Augmented Generation (RAG) techniques have exhibited remarkable performance across a wide ran

Ground-Level Near Real-Time Modeling for PM2.5 Pollution Prediction

arXiv:2604.18973v1 Announce Type: cross Abstract: Air pollution is a worldwide public health threat that can cause or exacerbate many illnesses, including respiratory disease, cardiovascular disease,

Guiding Distribution Matching Distillation with Gradient-Based Reinforcement Learning

arXiv:2604.19009v1 Announce Type: cross Abstract: Diffusion distillation, exemplified by Distribution Matching Distillation (DMD), has shown great promise in few-step generation but often sacrifices q

HALO: Hybrid Auto-encoded Locomotion with Learned Latent Dynamics, Poincare Maps, and Regions of Attraction

arXiv:2604.18887v1 Announce Type: new Abstract: Reduced-order models are powerful for analyzing and controlling high-dimensional dynamical systems. Yet constructing these models for complex hybrid sys

safetyarxiv-cs-ro

Hierarchically Robust Zero-shot Vision-language Models

arXiv:2604.18867v1 Announce Type: cross Abstract: Vision-Language Models (VLMs) can perform zero-shot classification but are susceptible to adversarial attacks. While robust fine-tuning improves their

How does the optimizer implicitly bias the model merging loss landscape?

arXiv:2510.04686v2 Announce Type: replace-cross Abstract: Model merging combines independent solutions with different capabilities into a single one while maintaining the same inference cost. Two popu

How to Teach Large Multimodal Models New Skills

arXiv:2510.08564v2 Announce Type: replace Abstract: How can we teach large multimodal models (LMMs) new skills without erasing prior abilities? We study sequential fine-tuning on five target skills wh

Hybrid Task and Motion Planning with Reactive Collision Handling for Multi-Robot Disassembly of Complex Products: Application to EV Batteries

arXiv:2509.21020v2 Announce Type: replace Abstract: This paper addresses the problem of multi-robot coordination for complex manipulation task sequences. We present a vision-driven task-and-motion pla

safetyarxiv-cs-ro

Integrating Anomaly Detection into Agentic AI for Proactive Risk Management in Human Activity

arXiv:2604.19538v1 Announce Type: new Abstract: Agentic AI, with goal-directed, proactive, and autonomous decision-making capabilities, offers a compelling opportunity to address movement-related risk

Investigating Counterfactual Unfairness in LLMs towards Identities through Humor

arXiv:2604.18729v1 Announce Type: new Abstract: Humor holds up a mirror to social perception: what we find funny often reflects who we are and how we judge others. When language models engage with hum

Knowledge-Guided Time-Varying Causal Inference for Arctic Sea Ice Dynamics

arXiv:2601.17647v2 Announce Type: replace-cross Abstract: Quantifying the causal relationship between sea ice thickness and sea surface height (SSH) is essential for understanding the mechanisms drivi

Large Language Models Exhibit Normative Conformity

arXiv:2604.19301v1 Announce Type: new Abstract: The conformity bias exhibited by large language models (LLMs) can pose a significant challenge to decision-making in LLM-based multi-agent systems (LLM-

LASER: Learning Active Sensing for Continuum Field Reconstruction

arXiv:2604.19355v1 Announce Type: cross Abstract: High-fidelity measurements of continuum physical fields are essential for scientific discovery and engineering design but remain challenging under spa

Learning Hybrid-Control Policies for High-Precision In-Contact Manipulation Under Uncertainty

arXiv:2604.19677v1 Announce Type: cross Abstract: Reinforcement learning-based control policies have been frequently demonstrated to be more effective than analytical techniques for many manipulation

Learning to Credit the Right Steps: Objective-aware Process Optimization for Visual Generation

arXiv:2604.19234v1 Announce Type: new Abstract: Reinforcement learning, particularly Group Relative Policy Optimization (GRPO), has emerged as an effective framework for post-training visual generativ

Let me say this clearly: LLMs cannot feel emotions. Emotions are evolutionary mechanisms. They push us to avoid danger or approach what is b…

Let me say this clearly: LLMs cannot feel emotions. Emotions are evolutionary mechanisms. They push us to avoid danger or approach what is beneficial. We experience emotions because we are alive, and

🦤 LeWorldModel: Learning Physics from Pixels — Stable World Models with Just Two Losses World models: 1️⃣ DINO-WM: pretrained ViT encoder (…

🦤 LeWorldModel: Learning Physics from Pixels — Stable World Models with Just Two Losses World models: 1️⃣ DINO-WM: pretrained ViT encoder (from ImageNet) → features → predictor. But encoder is frozen,

safetyyann-lecun--x

LiteParse: our open-source, layout-aware PDF parser for AI agents. The secret? Grid projection. Instead of heavy ML layout models or flat te…

LiteParse: our open-source, layout-aware PDF parser for AI agents. The secret? Grid projection. Instead of heavy ML layout models or flat text extraction, it projects text onto a monospace grid so ali

safetyjerry-liu--x

LLMs Know They're Wrong and Agree Anyway: The Shared Sycophancy-Lying Circuit

arXiv:2604.19117v1 Announce Type: new Abstract: When a language model agrees with a user's false belief, is it failing to detect the error, or noticing and agreeing anyway? We show the latter. Across

Location Not Found: Exposing Implicit Local and Global Biases in Multilingual LLMs

arXiv:2604.19292v1 Announce Type: cross Abstract: Multilingual large language models (LLMs) have minimized the fluency gap between languages. This advancement, however, exposes models to the risk of b

LogosKG: Hardware-Optimized Scalable and Interpretable Knowledge Graph Retrieval

arXiv:2604.18913v1 Announce Type: new Abstract: Knowledge graphs (KGs) are increasingly integrated with large language models (LLMs) to provide structured, verifiable reasoning. A core operation in th

Low-Rank Adaptation for Critic Learning in Off-Policy Reinforcement Learning

arXiv:2604.18978v1 Announce Type: cross Abstract: Scaling critic capacity is a promising direction for enhancing off-policy reinforcement learning (RL). However, larger critics are prone to overfittin

Lyapunov-Certified Direct Switching Theory for Q-Learning

arXiv:2604.19569v1 Announce Type: cross Abstract: Q-learning is one of the most fundamental algorithms in reinforcement learning. We analyze constant-stepsize Q-learning through a direct stochastic sw

M^{2}GRPO: Mamba-based Multi-Agent Group Relative Policy Optimization for Biomimetic Underwater Robots Pursuit

arXiv:2604.19404v1 Announce Type: cross Abstract: Traditional policy learning methods in cooperative pursuit face fundamental challenges in biomimetic underwater robots, where long-horizon decision ma