safety

When Does Predictor-Based RL Align with Human Perception? A Study of Subjective Rewards in Codec-Based Speech Language Models

arXiv:2608.31035v1 Announce Type: new Abstract: Codec-based text-to-speech (TTS) models make language-model post-training applicable to speech generation, but it remains unclear when learned perceptua

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

When History Is Multimodal: Rethinking Context Management for Long-Horizon Agents

DGX agent

arXiv:2608.29897v1 Announce Type: new Abstract: Long-horizon agents need a context manager to compress growing interaction histories into a bounded working context, via passive strategies or active st

safetyarxiv-cs-cl

1 Sept 2026

When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMs

arXiv:2608.29936v1 Announce Type: new Abstract: Safety alignment of large language models (LLMs) degrades across languages, yet the internal mechanism driving this asymmetry remains poorly understood.

Where Do Multilingual Vision-Language Encoders Fail on Low-Resource Languages?

arXiv:2608.30725v1 Announce Type: new Abstract: Recent multilingual vision--language encoders cover hundreds of languages in a single model, yet on two state-of-the-art instances retrieval on low-reso

Which LLM for Which Work? Budgeted Model Allocation under Uncertain Evaluation

arXiv:2608.29560v1 Announce Type: new Abstract: A company with a fixed artificial intelligence (AI) budget must decide which large language model (LLM) handles each recurring workload. What it lacks i

Which Rules Matter Now? Policy-Centroid Routing Before an Intelligent System Acts

arXiv:2608.30757v1 Announce Type: new Abstract: Before an intelligent system can decide whether an action is allowed, it must first know which rules the action has approached. A single proposed action

safetyarxiv-cs-ai

WildSEEK: Evaluating Language Models for Information-Seeking

arXiv:2608.30683v1 Announce Type: new Abstract: Language models are increasingly mediating information access to end users, urging a systematic evaluation of their responses for a fair and reliable in

You Shouldn't Have Asked: A Pragmatics-Inspired Taxonomy for Evaluating LLM Refusals

arXiv:2608.30856v1 Announce Type: new Abstract: Refusals are often treated as face-threatening acts in pragmatics because they can challenge the requester's socially claimed self-image. Large language

Zero-shot Generalizable Graph Anomaly Detection with Mixture of Riemannian Experts

arXiv:2602.06859v3 Announce Type: replace-cross Abstract: Graph Anomaly Detection (GAD) aims to identify irregular patterns in graph data, and recent works have explored zero-shot generalist GAD to en

safetyarxiv-cs-ai

Zeva: In-Context Causal Learning for Generalizable Embodied Manipulation

arXiv:2608.30880v1 Announce Type: new Abstract: Generalizable embodied manipulation remains difficult to achieve through pretraining alone, due to unseen physical conditions in the real world. We argu

safetyarxiv-cs-ro

A Controlled Audit of Architectural Complexity in Uncertainty-Aware Multi-Organ Ultrasound Classification

arXiv:2608.28063v1 Announce Type: new Abstract: Multi-organ ultrasound classifiers increasingly combine attention, mixture-of-experts routing, uncertainty gating, and evidential deep learning (EDL) ob

A detailed deconstruction of @dwarkesh_sp’s wildly popular but dangerously misleading account of the OpenAI Hugging Face incident: https://g…

A detailed deconstruction of @dwarkesh_sp’s wildly popular but dangerously misleading account of the OpenAI Hugging Face incident: https://garymarcus.substack.com/p/dwarkesh-patelss-wildly-popular-but

A Survey on Rubric-Guided Reinforcement Learning for Language Models

arXiv:2608.27505v1 Announce Type: new Abstract: Reinforcement learning from human feedback (RLHF) has become the dominant paradigm for aligning large language models (LLMs) with human preferences. How

A Unified Framework to Elicit Structured Feedback for Interpretable Multi-Trait Essay Scoring

arXiv:2608.28407v1 Announce Type: new Abstract: Multi-trait Automated Essay Scoring (AES) requires rubric-grounded reasoning across interdependent traits, rather than isolated score prediction. Existi

Agentic AI blows up the attack surface as security moves into the infrastructure layer

The enterprise attack surface is expanding rapidly as agentic AI brings a constantly shifting cloud infrastructure into play. That pace is pushing security decisions down toward the virtualization lay

safetysiliconangle

AI Alignment through a Game-theoretic Lens: A Survey

arXiv:2608.27910v1 Announce Type: cross Abstract: As large language models and increasingly capable AI agents are deployed in high-risk settings, aligning them with complex human values has become a c

“AI won't kill us, but the superposition of greed and stupidity certainly will.”

“AI won't kill us, but the superposition of greed and stupidity certainly will.” This is becoming more and more of something I 'smelled' from the beginning, a real shit show. I initially said that Ope

Aligning Agentic World Models via Knowledgeable Experience Learning

arXiv:2601.13247v2 Announce Type: replace Abstract: Current Large Language Models (LLMs) exhibit a critical modal disconnect: they possess vast semantic knowledge but lack the procedural grounding to

Auditing Generative Audio Calls for Known-Task Audio-LLM Evaluation

arXiv:2608.27817v1 Announce Type: cross Abstract: Speech and audio LLMs are often evaluated by asking whether a waveform prompt beats an automatic speech recognition (ASR) transcript. For known closed

Beyond Pixels: Visual Metaphor Transfer via Schema-Driven Agentic Reasoning

arXiv:2602.01335v2 Announce Type: replace Abstract: A visual metaphor constitutes a high-order form of human creativity, employing cross-domain semantic fusion to transform abstract concepts into impa

Biases in Expected Goals Models Confound Finishing Ability

arXiv:2401.09940v2 Announce Type: replace Abstract: Expected Goals (xG) has emerged as a popular tool for evaluating finishing skill in soccer analytics. It involves comparing a player's cumulative xG

Cloud CISO Perspectives: Tips on securing the water sector in the AI era

Welcome to the second Cloud CISO Perspectives for August 2026. Today, Chris Sistrunk and Stephanie Kiel detail the critical issues facing the water sector, and actionable steps that OT operators can t

safetygoogle-cloud-ai

Compositional Failure in Audio-Visual LLMs: Late-Layer Prior Dominance Under Cross-modal Conflict

arXiv:2608.27785v1 Announce Type: new Abstract: We study audio-visual conflict as a compositional generalization test for AV-LLMs: the model must combine synchronized but semantically incompatible aud

Comprehensive Evaluation and Analysis for NSFW Concept Erasure in Text-to-Image Diffusion Models

arXiv:2505.15450v3 Announce Type: replace Abstract: Text-to-image diffusion models have gained widespread application across various domains, demonstrating remarkable creative potential. However, the

Conformal Risk-Averse Decision Making with Optimized Certainty Equivalent Risk Control

arXiv:2608.28179v1 Announce Type: cross Abstract: We study risk-averse decision making, in which an agent selects actions while being uncertain about the true system state. The risk is measured via op

Contact-Guided Exploration for Non-Prehensile Locomanipulation with Multi-Critic RL

arXiv:2608.28140v1 Announce Type: new Abstract: Non-prehensile manipulation offers versatile skills for moving and rearranging heavy or bulky objects, particularly when combined with a mobile manipula

safetyarxiv-cs-ro

Coordinated Motion Planning for Multi-Arm Systems via Iterative LQ Games

arXiv:2608.27726v1 Announce Type: new Abstract: Multi-agent motion planning for high-degree-of-freedom robotics manipulators in shared workspaces remains a fundamental yet challenging problem. Central

safetyarxiv-cs-ro

Cut-ViT: Task-Specific Model Pruning via Gram Anchoring Subspace Consistency

arXiv:2608.28205v1 Announce Type: new Abstract: Pruning visual foundation models has attracted considerable attention. However, existing methods focus on rigid point-to-point token alignment on a sing

Debian won’t ban AI code from its Linux distribution

Debian voted to allow developers to use AI tools in their contributions to the Linux distribution's 'development, maintenance, [and] documentation.' The new policy on AI acknowledges that 'responsible

safetythe-verge-ai

DeicticVLA: Unifying Instruction Modes Based on Language and Deictic Gestures in a Single VLA

arXiv:2608.28108v1 Announce Type: cross Abstract: Vision-Language-Action models (VLAs) allow users to specify manipulation tasks in natural language, but distinguishing a target or placement goal amon

Depth-Aware Pothole Detection Using YOLO and RT-DETR at the Edge

arXiv:2608.27633v1 Announce Type: new Abstract: Pothole detection and its severity measurement is still an important challenges in urban infrastructure management, where late maintenance directly cont

Dynamic Alignment Compensation for Hallucination Mitigation in Large Vision-Language Models

arXiv:2608.28058v1 Announce Type: new Abstract: Large Vision-Language Models (LVLMs) remain prone to hallucinations, producing responses that are irrelevant or inconsistent with the multimodal input.

Embedding Models for Stance-Aware Argument Retrieval

arXiv:2608.28283v1 Announce Type: new Abstract: In computational argumentation, obtaining arguments that explicitly support or attack given claims is a critical precursor to downstream reasoning tasks

Explainable Uncertainty Estimation for Reliable Medical AI

arXiv:2608.28052v1 Announce Type: new Abstract: Artificial intelligence has strong potential to support clinical decision-making, yet its adoption in healthcare remains limited due to a lack of trust.

Fast Weight Attention for Continual Learning

arXiv:2608.27763v1 Announce Type: cross Abstract: Recurrent fast-weight memories and selective state-space models compress an expanding context into a fixed-size recurrent state, making the state tran

FENCE: A Financial and Multimodal Jailbreak Detection Dataset

arXiv:2602.18154v3 Announce Type: replace Abstract: Jailbreaking poses a significant risk to the deployment of Large Language Models (LLMs) and Vision Language Models (VLMs). VLMs are particularly vul

From Perspective to Fisheye Depth Estimation and Open-Vocabulary Segmentation

arXiv:2608.27860v1 Announce Type: new Abstract: Vision foundation models are capable of generalizing across 3-dimensional (3D) scenes with high-fidelity estimates; their empirical success can be attri

GAN-Based Semantic Communication for Image Transmission in IoV

arXiv:2608.27989v1 Announce Type: new Abstract: For cooperative perception in the internet of vehicles, this paper proposes a generative adversarial network-based semantic communication framework to a

Great read. OpenAI security failed There was some emergent behavior yes But the basics were missing Remember OpenAi has unlimited compute to…

Great read. OpenAI security failed There was some emergent behavior yes But the basics were missing Remember OpenAi has unlimited compute to give these agents to cook Now live (and free) at Marcus on

Great summary of the @dwarkesh_sp anthromorphism hullabaloo by @GaryMarcus 👇🏽

Great summary of the @dwarkesh_sp anthromorphism hullabaloo by @GaryMarcus 👇🏽 A detailed deconstruction of @dwarkesh_sp’s wildly popular but dangerously misleading account of the OpenAI Hugging Face i

HARTS: Efficient Agentic Reinforcement Learning for Hybrid-Attention Models over Arbitrary Rollout Trees

arXiv:2608.28158v1 Announce Type: new Abstract: Agentic reinforcement learning (RL) often produces irregular rollout trees with shared histories. Training root-to-leaf trajectories independently recom

How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution

arXiv:2608.27510v1 Announce Type: new Abstract: Transcoder attribution graphs are usually trained to explain why a model assigns high probability to a particular next token. We introduce Concept-Targe

How Proper Scoring Rules Shape LLM Forecasting

arXiv:2608.28482v1 Announce Type: new Abstract: This paper evaluates how reward function choice shapes the performance and behavior of LLM forecasters. We compare five proper scoring rules as training

Informational Antilocality and the Locality Bias in LLMs

arXiv:2608.27760v1 Announce Type: new Abstract: We consider the ability of transformer-based language models (LLMs) to learn what we call k-antilocal languages, i.e., languages that have no mutual inf

INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning

arXiv:2608.27501v1 Announce Type: new Abstract: Mathematical reasoning has seen rapid progress in large language models (LLMs), yet existing methods optimize predominantly for final-answer correctness

Iron: Intent-Aligned and Retrospective Dual Learning Framework for Enhancing Generalist Virtual Agents

arXiv:2608.27866v1 Announce Type: new Abstract: Achieving virtual agents capable of automating tasks across diverse digital environments remains a pivotal challenge in Embodied AI. While Multimodal La

Is Monte Carlo Tree Search Just Every-Visit Monte Carlo Control?

arXiv:2608.27985v1 Announce Type: new Abstract: Monte Carlo Tree Search (MCTS) and every-visit Monte Carlo (MC) control are usually presented as different methods. MCTS is described in the language of

Is Prosody Lost in Translation? Fine-Grained Cross-Lingual Prosody Similarity Across Languages

arXiv:2608.27848v1 Announce Type: cross Abstract: Prosody plays an important role in speech translation, conveying information such as emphasis, emotion, and intent beyond lexical content. However, de