Search: “safety”

Rethinking Feature Alignment in Generalist Graph Anomaly Detection: A Relational Fingerprint-based Approach

arXiv:2605.25429v1 Announce Type: new Abstract: Generalist graph anomaly detection (GAD) aims to detect anomalies on unseen graphs without graph-specific retraining. Nevertheless, existing approaches

Safety

Rewarding Structural Conformance of Reasoning using Process Mining

DGX agent

arXiv:2510.25065v3 Announce Type: replace Abstract: Recent advances in sparse reward policy gradient methods have enabled effective reinforcement learning (RL)-based language model post-training. Howe

safetyarxiv-cs-ai

26 May 2026

Right-Sizing Communication and Recommendation Set Size in AI-Assisted Search

arXiv:2605.23944v1 Announce Type: new Abstract: We model the interaction between a user and an AI driven recommendation system. The user initiates the process by conveying preference information throu

RiskBridge: Turning CVEs into Business-Aligned Patch Priorities

arXiv:2601.06201v2 Announce Type: replace-cross Abstract: Enterprises are confronted with an unprecedented escalation in cybersecurity vulnerabilities, with thousands of new CVEs disclosed each month.

SEAL: Synergistic Co-Evolution of Agents and Learning Environments

arXiv:2605.24426v1 Announce Type: new Abstract: Large Language Model (LLM) agents are increasingly improved through interaction, yet most self-evolution methods adapt either the policy or the learning

Selective Latent Thinking: Adaptive Compression of LLM Reasoning Chains

arXiv:2605.25745v1 Announce Type: new Abstract: Explicit chain-of-thought (CoT) reasoning substantially improves the reasoning ability of large language models (LLMs), but incurs high inference cost d

SliceWorld: A Predictive and Controllable World-State Model for CT Report Generation

arXiv:2605.24371v1 Announce Type: cross Abstract: CT report generation (CTRG) requires models to summarize three-dimensional anatomical context and pathological findings from hundreds of axial slices.

Smoother Action Chunking Flow Policy via Prior-Corrected Orthogonal Trust-Region Guidance

arXiv:2605.24433v1 Announce Type: cross Abstract: Flow-matching robot policies commonly use action-chunking inference for efficient closed-loop control, but chunk boundaries can introduce discontinuou

Model Releases

SoK: A Comprehensive Security Analysis of Jailbreak Resilience in GPT and DeepSeek Models

model-releasesarxiv-cs-ai

arXiv:2506.18543v2 Announce Type: replace-cross Abstract: The rapid proliferation of Large Language Models (LLMs) has heightened concerns regarding their exposure to jailbreak attacks, which craft adv

SpaceX has lost 13B since 2023 and #WallStreet is still pricing the IPO at 1 TRILLION. Though a mere economist, @deanbaker13 says the math…

SpaceX has lost 13B since 2023 and #WallStreet is still pricing the IPO at 1 TRILLION. Though a mere economist, @deanbaker13 says the math isn't adding up. https://cepr.net/publications/wall-street-sa

SpaceX’s unconventional corporate arrangements appear to benefit Elon Musk at the expense of other shareholders, experts said. https://nyti.…

SpaceX's corporate structure and financial arrangements have been criticized by experts as potentially favoring Elon Musk's interests over those of other shareholders. The article examines how the com

SpecAlign: A Semantic Alignment Framework for SystemVerilog Assertion Generation

arXiv:2605.25181v1 Announce Type: new Abstract: Existing Large Language Model (LLM) approaches to SystemVerilog Assertion (SVA) generation primarily focus on syntactic validity and formal verification

StakeBench: Evaluating Language Understanding Grounded in Market Commitment

arXiv:2605.26074v1 Announce Type: cross Abstract: Existing financial NLP benchmarks often rely on labels supplied by outside observers, measuring how language is perceived rather than what speakers ha

STaT: Resolving Shape Distortion in Non-Stationary Time Series via Tri-Modal Synergy

arXiv:2605.25943v1 Announce Type: new Abstract: Recent research in time series forecasting frequently investigates the integration of textual and visual modalities with numerical models to better navi

Stop Comparing LLM Agents Without Disclosing the Harness

arXiv:2605.23950v1 Announce Type: new Abstract: This position paper argues that, for long-horizon tasks evaluated across models with comparable frontier capability, the agent execution harness, namely

Strat-Reasoner: Reinforcing Strategic Reasoning of LLMs in Multi-Agent Games

arXiv:2605.04906v2 Announce Type: replace Abstract: While Large Language Models (LLMs) excel in certain reasoning tasks, they struggle in multi-agent games where the final outcome depends on the joint

Subspace-Guided Semantic and Topological Invariant Registration for Annotation-Free Ultrasound Plane Quality Control

arXiv:2605.25396v1 Announce Type: cross Abstract: Reliable quality control (QC) of ultrasound images is essential for both real-time acquisition guidance and retrospective clinical audit, yet existing

Summoning the Oracle to Slay It: Mitigating Look-Ahead Bias in Financial Backtesting with Large Language Models

arXiv:2605.24564v1 Announce Type: new Abstract: Backtesting large language models (LLMs) on historical financial data is unreliable because pre-training cuts off after the events happened. An LLM trai

TapSampling: Inference-Time Sampling with a Task-Progress-Understanding Verifier for Robotic Manipulation

arXiv:2605.25547v1 Announce Type: new Abstract: Existing embodied control research demonstrates remarkable performance improvements by scaling training data and model size. We instead explore inferenc

safetyarxiv-cs-ro

Task-Aligned Self-Supervised Learning for Medical Image Analysis: A Systematic Review and Practical Design Guidelines

arXiv:2605.23995v1 Announce Type: cross Abstract: Self-supervised learning (SSL) has emerged as a promising paradigm for addressing the annotation bottleneck in medical imaging by learning representat

temporarily putting PhD in my social media name because this tweet from Elon is so asinine and because Elon’s goons went after a friend for …

temporarily putting PhD in my social media name because this tweet from Elon is so asinine and because Elon’s goons went after a friend for supporting the poor kid that Elon so rudely attacked. @iScie

The Behavioral Credibility Trilemma: When Calibrated Autonomy Becomes Impossible

arXiv:2605.25739v1 Announce Type: new Abstract: We prove that no reinforcement learning policy with confidence-gated autonomy can simultaneously achieve maximum helpfulness, optimal calibration, and f

The Concept Allocation Zone: Tracking How Concepts Form Across Transformer Depth

arXiv:2605.24856v1 Announce Type: cross Abstract: Concept formation in transformer language models is depth-extended, not a single-layer event: concepts emerge gradually across a contiguous region of

The Implicit Bias of Adam and Muon on Smooth Homogeneous Neural Networks

arXiv:2602.16340v3 Announce Type: replace Abstract: We study the implicit bias of momentum-based optimizers on smooth homogeneous models. We show that extit{momentum steepest descent} algorithms like

The main thing that come through reading the SpaceX S-1 is how much Musk and Altman have in common.

This post draws a comparison between Elon Musk and Sam Altman based on insights from SpaceX's S-1 filing, highlighting shared characteristics or philosophies between the two tech leaders. The analysis

The Many Faces of On-Policy Distillation: Pitfalls, Mechanisms, and Fixes

arXiv:2605.11182v2 Announce Type: replace Abstract: On-policy distillation (OPD) and on-policy self-distillation (OPSD) have emerged as promising post-training methods for large language models, offer

The Path Matters: Learning a Token-Commitment Policy for Diffusion Language Models

arXiv:2605.24697v1 Announce Type: cross Abstract: Diffusion large language models promise faster generation by refining many token positions in parallel, but this parallelism introduces a hidden contr

The SpaceX IPO is a shitshow. When you peel back the layers, you realize how thoroughly corrupt it is.

The SpaceX IPO is a shitshow. When you peel back the layers, you realize how thoroughly corrupt it is. SpaceX’s unconventional corporate arrangements appear to benefit Elon Musk at the expense of othe

TopoAlign: Topology-Aware Visual Representation Alignment

arXiv:2605.25541v1 Announce Type: cross Abstract: Neural networks encode inputs as high-dimensional vectors, known as representations, that capture how models process data by encoding task-relevant st

Toward Reliable Design of LLM-Enabled Agentic Workflows: Optimizing Latency-Reliability-Cost Tradeoffs

arXiv:2605.23929v1 Announce Type: new Abstract: Modern AI systems increasingly rely on workflows composed of multiple interacting agents, some powered by large language models (LLMs) and others by con

Towards Cognitively-Faithful Decision-Making Models to Improve AI Alignment

arXiv:2509.04445v2 Announce Type: replace Abstract: Recent AI trends seek to align AI models to learned human-centric objectives, such as personal preferences, utility, or societal values. Using stand

Towards Inclusive Toxic Content Moderation: Addressing Vulnerabilities to Adversarial Attacks in Toxicity Classifiers Tackling LLM-generated Content

arXiv:2509.12672v2 Announce Type: replace Abstract: The volume of machine-generated content online has grown dramatically due to the widespread use of Large Language Models (LLMs), leading to new chal

Towards Low-Gravity Planetary Exploration using Reinforcement Learning for Walking, Jumping, and In-flight Attitude Control

arXiv:2605.24643v1 Announce Type: new Abstract: This paper presents reinforcement learning (RL) policies for dynamic quadrupedal locomotion in planetary exploration scenarios. Building on a taskoptimi

safetyarxiv-cs-ro

Towards the Connection between Activation Sparsity and Flat Minima

arXiv:2605.25612v1 Announce Type: cross Abstract: The observation that activation sparsity emerges in MLP blocks of standardly trained Transformers offers an opportunity to drastically reduce computat

Towards Understanding Adam Convergence on Highly Degenerate Polynomials

arXiv:2603.09581v2 Announce Type: replace Abstract: Adam is a widely used optimization algorithm in deep learning, yet the specific class of objective functions where it exhibits inherent advantages r

Trait-Aware Policy Optimization for Autoregressive Multi-Trait Essay Scoring

arXiv:2605.25731v1 Announce Type: new Abstract: Multi-trait essay scoring aims to provide fine-grained evaluation of writing quality across multiple dimensions. However, how to effectively post-train

Trust-Aware Joint Feature-Prediction Discrepancy for Robust Domain Adaptation

arXiv:2605.25119v1 Announce Type: cross Abstract: Domain adaptation aims to mitigate performance degradation caused by distribution shifts between a labeled source domain and an unlabeled or sparsely

Uncertainty-DTW for Sequences and Visual Tokens

arXiv:2605.25110v1 Announce Type: cross Abstract: Aligning structured data is a fundamental problem in computer vision and machine learning, underlying tasks such as time series analysis, human action

Unifying Value Alignment and Assignment in Cross-Domain Offline Reinforcement Learning with Heterogeneous Datasets

arXiv:2605.24862v1 Announce Type: new Abstract: Cross-domain offline reinforcement learning (RL) aims to learn a policy in the target domain with a limited target domain dataset and a source domain da

UniRank: End-to-End Domain-Specific Reranking of Hybrid Text-Image Candidates

arXiv:2603.29897v2 Announce Type: replace-cross Abstract: Reranking is a critical component in many information retrieval pipelines. Despite remarkable progress in text-only settings, multimodal reran

Universal Boosts, Specific Suppressors: Sparse Autoencoder Steering of Medical Vision-Language Models

arXiv:2605.24977v1 Announce Type: cross Abstract: Medical vision-language models (VLMs) often hallucinate findings when generating chest X-ray reports: they fabricate findings that are not present in

University of California STEM professors want standardized tests back due to severe math deficiencies among students: “We now observe prepar…

University of California STEM professors want standardized tests back due to severe math deficiencies among students: “We now observe preparation gaps so severe that instructors must reteach middle sc

“Unserious, empty, hallucinatory, and borderline dishonest” - the prospectus for a company that the S&P 500 is about to jam down your throat…

Gary Marcus criticizes a company's prospectus as containing unserious, empty, and potentially dishonest claims before its inclusion in the S&P 500 index. The post appears to highlight concerns about i

VEN-VL: A Visual Ensemble MoE Framework for Effective and Efficient Multi-Modal Understanding

arXiv:2605.25952v1 Announce Type: cross Abstract: Despite the remarkable progress achieved by recent efficient methods in accelerating multimodal understanding, they still suffer from noticeable perfo

Vision-Guided Outdoor Flight and Obstacle Evasion via Reinforcement Learning

arXiv:2605.24449v1 Announce Type: cross Abstract: Although quadcopters boast impressive traversal capabilities enabled by their omnidirectional maneuverability, the need for continuous pilot control i

What Gets Cited: Competitive GEO in AI Answer Engines

arXiv:2605.25517v1 Announce Type: new Abstract: AI answer engines generate answers from retrieved pages but cite only a few sources. This makes visibility depend not just on ranking, but on being cite

When Does Multi-Agent RL Improve LLM Workflows? Workflow, Scale, and Policy-Sharing Tradeoffs

arXiv:2605.24202v1 Announce Type: new Abstract: Multi-agent LLM workflows route inference through specialized roles to lift end-task accuracy, but jointly training those roles with reinforcement learn

When Self-Belief Misleads: Active Label Acquisition for Reinforcement Learning with Verifiable Rewards

arXiv:2605.25864v1 Announce Type: cross Abstract: Large Language Models (LLMs) have achieved remarkable advancements in reasoning capabilities empowered by Reinforcement Learning with Verifiable Rewar