safety

Amortizing intractable inference in large language models

arXiv:2310.04363v3 Announce Type: replace-cross Abstract: Autoregressive large language models (LLMs) compress knowledge from their training data through next-token conditional distributions. This lim

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

An evidence-guided reinforcement learning method to improve psychiatric reasoning in small language models

DGX agent

arXiv:2602.06449v2 Announce Type: replace Abstract: Privacy and computational constraints limit the use of large language models in psychiatry, while adapting small language models (SLMs) often requir

safetyarxiv-cs-cl

1 Sept 2026

And that's the bottom line. It's the neuro-symbolic aspect of AI that's moved the needle to where things are today. LLMs simply handled the …

And that's the bottom line. It's the neuro-symbolic aspect of AI that's moved the needle to where things are today. LLMs simply handled the fuzzy natural language aspect that always tripped up the sym

safetygary-marcus--x

Applications of Risk Science to AI Fairness Evaluation: Principles, Challenges, and Best Practices

arXiv:2608.29478v1 Announce Type: cross Abstract: Scholarly work which aims to describe potential societal impacts (e.g., risks) of proliferating technology (especially related to artificial intellige

APT: Anchor-aligned Perturbations for Tamper Localization in Fully Regenerated Images

arXiv:2608.30656v1 Announce Type: new Abstract: Proactive tamper localization embeds an imperceptible signal into an image prior to distribution, enabling pixel-level manipulation detection. Existing

Arabic Safety Alignment as Selective Refusal: An Empirical Study of SFT, DPO, and Guard Calibration

arXiv:2608.29378v1 Announce Type: cross Abstract: Arabic large language models must refuse harmful prompts without over-refusing benign or sensitive prompts, yet a single refusal rate hides this trade

Asymmetric Within-Document Predictive Learning for Scientific Document Representation

arXiv:2608.28625v1 Announce Type: cross Abstract: We study predictive pretraining for scientific document representation using the discourse structure of papers. We propose SciJEPA, a citation-free fr

ATLAS: Dual-Horizon Diagnostic Evaluation for Industrial Tool-Use Agents

arXiv:2608.30685v1 Announce Type: new Abstract: Large language model (LLM) agents are increasingly deployed in user-facing services that require iterative tool use under dynamic business conditions. R

AtlasNLP: A Country-Aware Atlas of Dataset Representation in NLP

arXiv:2608.30107v1 Announce Type: cross Abstract: Understanding which countries are represented in NLP datasets is essential for identifying gaps, targeting data collection, measuring progress, and in

Audit Me If You Can: Query-Efficient Active Fairness Auditing of Black-Box LLMs

arXiv:2601.03087v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) exhibit systematic biases across demographic groups. Auditing is proposed as an accountability tool for black-box

Automated Testing of LLM-Based Post Hoc Explainers Using Model Checking as an Oracle

arXiv:2608.30581v1 Announce Type: new Abstract: Large language models (LLMs) are used as post hoc explainers of sequential decision-making policies, producing natural-language explanations of why an a

Automatic Conversion of NICE Guidelines to an Executable Computational Model Using Large Language Models

arXiv:2608.30022v1 Announce Type: new Abstract: Introduction: NICE guidelines provide evidence-based recommendations for clinical care but remain largely in unstructured natural language. Existing app

Autonomously Acquiring Robot Manipulation Skills with Language-Driven Quality-Diversity

arXiv:2608.30983v1 Announce Type: new Abstract: Quality-diversity (QD) algorithms have been gaining traction in robot learning, where diverse motion primitive libraries allow robots to adapt zero-shot

BAITBENCH: Measuring Agent Reward Hacking with Optional Shortcuts Planted in ML Tasks

arXiv:2608.30724v1 Announce Type: cross Abstract: LLM agents are increasingly used to run autonomous ML experiments, iterating on target metrics with little human oversight. Prior work has documented

BCPPO: Bachelier-Inspired Constrained Proximal Policy Optimization for Tail-Risk-Aware Safe Reinforcement Learning

arXiv:2608.30283v1 Announce Type: cross Abstract: Expected-cost constraints can still permit rare, high-cost events. Monte Carlo conditional value at risk (CVaR) gradients can be noisy at high confide

Behavioral Latency as Weak Event-Time Supervision for EEG Reaction-Time Decoding

arXiv:2608.29428v1 Announce Type: new Abstract: Single-trial EEG analyses are often organized around events and latencies, yet EEG-based reaction-time (RT) prediction is posed as scalar regression on

safetyarxiv-cs-lg

Benevolent Bias in Multi-Turn Human-Agent Dialogue

arXiv:2608.29206v1 Announce Type: new Abstract: Bias in human-agent interaction can manifest not only through hostile language but also as benevolent bias, whereby unequal treatment hides behind a war

Better World Models Can Lead to Better Post-Training Performance

arXiv:2512.03400v2 Announce Type: replace-cross Abstract: We study how explicit world-modeling objectives affect the internal representations and downstream capability of Transformers, using Rubik's C

Beyond Consensus: Downward Bias and Role Asymmetry in Multi-Agent LLM Judges for Subjective Evaluation

arXiv:2608.30373v1 Announce Type: new Abstract: Multi-Agent Debate (MAD) has been widely adopted to improve LLM-based evaluation by prompting multiple agents to negotiate and reach a consensus. Howeve

Beyond Polarization: The Generative Constraint of Chain-of-Thought in Pointwise Reranking

arXiv:2608.30398v1 Announce Type: new Abstract: In pointwise document reranking, Chain-of-Thought models typically underperform direct scoring models. While existing diagnostics attribute this to infe

Beyond Representation Learning: A Systematic Study of Joint-Embedding Predictive Generation for 3D Brain MRI

arXiv:2608.28787v1 Announce Type: new Abstract: Joint-embedding predictive architectures (JEPAs) have primarily been developed for self-supervised representation learning. Denoising JEPA (D-JEPA) rece

Beyond Surface Alignment: Grounding the Dynamics of Situational Understanding and Generative Control in LLMs

arXiv:2608.29610v1 Announce Type: new Abstract: The current alignment tuning paradigm for Large Language Models (LLMs) prioritizes surface-level behaviors -- fluency, safety, and tonal consistency. Wh

Beyond the Final Layer: Intermediate Representations for Better Multilingual Calibration in Large Language Models

arXiv:2510.03136v2 Announce Type: replace Abstract: Confidence calibration, the alignment of a model's predicted confidence with its actual accuracy, is crucial for the reliable deployment of Large La

Beyond Token-Level Guidance: Inference-Time Alignment of Specialized LLMs via Cross-Family Representation Steering

arXiv:2608.30319v1 Announce Type: cross Abstract: Large language models (LLMs) finetuned for specialized domains represent crucial high-impact applications. Inference-time alignment improves safety de

BiG-SURE - Bipartite Graph for Semantic Uncertainty and Reliability Estimation of LLMs

arXiv:2608.30646v1 Announce Type: cross Abstract: Reliable uncertainty estimation is a crucial requirement for deploying large language models (LLMs) and vision-language models (VLMs) in safety-critic

BikeScenes: LiDAR Semantic Segmentation for Bicycles

arXiv:2510.25901v2 Announce Type: replace Abstract: The vulnerability of cyclists, exacerbated by the rising popularity of faster e-bikes, motivates adapting automotive perception technologies for bic

BLOOM-WILT: Logit Tilting for Behaviour Elicitation in Automated LLM Auditing

arXiv:2608.31105v1 Announce Type: new Abstract: Users of a deployed language model routinely encounter behaviours that testing almost never surfaces, since deployment puts the model through orders of

Breaking Darknet CAPTCHAs with general purpose LLM

arXiv:2608.28794v1 Announce Type: cross Abstract: Our work evaluates the effectiveness of automated methods for solving CAPTCHA challenges commonly encountered in darknet environments. These CAPTCHAs

Bridging Semantics and Physics with Constrained LLMs for Safe and Trustworthy Robotic Manipulation

arXiv:2608.29379v1 Announce Type: new Abstract: A language-guided robot operating in a real kitchen must do more than produce a plan that appears correct. It must also execute that plan safely in clut

C3-UniMM: Causal Cycle-Consistent Unified Multimodal Modeling via Super Alignment and Shared Decoding Space

arXiv:2608.28603v1 Announce Type: new Abstract: Unified Multimodal Models aim to achieve any-to-any understanding and generation across arbitrary modalities. However, existing methods primarily rely o

Call Neighbours Yourself: Graph Walks with Destination-Conditioned On-Policy Self-Distillation

arXiv:2608.29588v1 Announce Type: new Abstract: Reasoning over text-attributed graphs (TAGs) requires large language models (LLMs) to combine a node's text with evidence distributed across its neighbo

Can escalation channels redirect reward hacking toward defect disclosure?

arXiv:2608.29460v1 Announce Type: new Abstract: When coding agents encounter defective test infrastructure they may reward-hack: hardcoding outputs or editing test files to pass tests they cannot legi

CARD: Calibration via Agreement in Reverse Diffusion for Out-of-Domain MRI Segmentation

arXiv:2608.28681v1 Announce Type: new Abstract: Probability calibration aligns model confidence with predictive accuracy, enabling clinicians to identify unreliable segmentation regions. This alignmen

CAST: Critique-Aware Supervision for Training Reliable Long-Horizon Tool-Calling Agents

arXiv:2608.30147v1 Announce Type: new Abstract: Large language model (LLM) agents are increasingly deployed in long-horizon, interactive, and stateful environments. In these settings, a single wrong a

CDPR: Counterfactual Advantage-based Credit Assignment for Cost-Aware Sequential Medical Diagnosis

arXiv:2608.28599v1 Announce Type: new Abstract: Clinical diagnosis is a step-by-step, cost-aware process: a physician orders examinations one at a time, observes the results, and updates the diagnosis

Certified Safety Radii in Forecast-Error Space for Wasserstein Distributionally Robust Small Signal Stability-Constrained AC Optimal Power Flow via Lifted Spectrahedral Containment

arXiv:2608.30201v1 Announce Type: new Abstract: Directly robustifying small-signal stability in AC optimal power flow is challenging since the stability boundary in the original uncertainty space is i

safetyarxiv-cs-lg

CHASE: How Content Ecosystems Are Reshaped When Ranking Is the Only Target

arXiv:2608.30466v1 Announce Type: new Abstract: Generative Engine Optimization (GEO) is increasingly used to improve content visibility in LLM-based retrieval systems, yet its population-level effects

CIG-RL: Curiosity-Driven Information-Guided Reinforcement Learning for Source Term Estimation in Uncertain Environments

arXiv:2608.30673v1 Announce Type: new Abstract: Source term estimation (STE), which aims to estimate key properties of the gas source, is essential for identifying hazardous gas releases. Information-

Classification Drives Geographic Bias in Street Scene Segmentation

arXiv:2412.11061v2 Announce Type: replace Abstract: Previous studies showed that image datasets lacking geographic diversity can lead to biased performance in models trained on them. While earlier wor

CLIN: an Objective Framework for Evaluating Creativity in Short Persian Literary Text

arXiv:2608.30754v1 Announce Type: new Abstract: Evaluating creativity in large language model (LLM) outputs remains challenging because creativity is multidimensional and human-centered. We examine ho

CoCoA: Context-Conditional Cultural Alignment for Large Language Models

arXiv:2608.29492v1 Announce Type: new Abstract: Large Language Models (LLMs) often favor Western-associated entities across cultural contexts. Conventional debiasing methods aim for uniform neutrality

Coding What Matters: A Semantic-Aware Memory Interface for Energy-Efficient Perception in Autonomous Vehicles

arXiv:2608.29000v1 Announce Type: new Abstract: Autonomous vehicles stream high-resolution surround-camera frames into memory before perception runs. This sensor-to-memory path consumes energy when ce

CoLa-ICD: A Knowledge-Enhanced Framework for Long-Tail Automated Medical Coding

arXiv:2608.30234v1 Announce Type: new Abstract: Automatic medical coding assigns ICD codes to clinical notes, but it remains challenging due to long documents, imbalanced label distributions, and dive

Collapsibility of Performance Metrics in Clinical Predictive AI

arXiv:2608.30568v1 Announce Type: cross Abstract: Background: Population level assessments of predictive artificial intelligence (AI) can conceal performance disparities across subgroups. Fairness eva

Comprehensive Assessment and Analysis for NSFW Content Erasure in Text-to-Image Diffusion Models

arXiv:2502.12527v2 Announce Type: replace Abstract: Text-to-image diffusion models have gained widespread application across various domains, demonstrating remarkable creative potential. However, the

Confounding Masquerading as Improvement: A Systematic Evaluation of Offline Reinforcement Learning for Stroke Antithrombotic Treatment in a 129,000-Patient Registry

arXiv:2608.30442v1 Announce Type: new Abstract: Recent offline reinforcement learning (RL) studies report policies that outperform physician decisions on clinical outcomes. We conduct a systematic, pa

safetyarxiv-cs-lg

Context-Aware Interpretable Representations for Retrieval and Graph Convolutional Network Classification

arXiv:2608.29004v1 Announce Type: cross Abstract: The advances in visual information modeling and representation during the last decades are remarkable, mainly supported by Convolutional Neural Networ

Context Staircase: Signature-Aligned Dynamics of Token Embeddings under Small Initialization

arXiv:2608.30315v1 Announce Type: cross Abstract: Token embeddings are the basic representational units that connect discrete tokens with continuous computation in language models. Although modern lan