safety

Rooted Absorbed Prefix Trajectory Balance with Submodular Replay for GFlowNet Training

arXiv:2603.00454v2 Announce Type: replace-cross Abstract: Generative Flow Networks (GFlowNets) enable fine-tuning large language models to approximate reward-proportional posteriors, but they remain p

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains

DGX agent

arXiv:2605.29156v1 Announce Type: cross Abstract: Pointwise reward modeling offers critical signals for LLM post-training, yet struggles with absolute scoring in subjective, non-verifiable settings. R

safetyarxiv-cs-cl

29 May 2026

Rubric-Guided Process Reward for Stepwise Model Routing

arXiv:2605.29310v1 Announce Type: new Abstract: Stepwise model routing improves the efficiency of Large Reasoning Models (LRMs) by assigning each reasoning step to a suitable model. Recent methods for

SafeRx-Agent: A Knowledge-Grounded Multi-Agent Framework for Safe and Explainable Medication Recommendation

arXiv:2605.29146v1 Announce Type: cross Abstract: Medication recommendation predicts medications for patient visits, but existing methods still face two key challenges. At the model level, traditional

SAHG: Sector-Anisotropic Hyperbolic Graph Model for Social Bot Detection

arXiv:2605.30166v1 Announce Type: cross Abstract: LLM-driven social bots can generate fluent, human-like text, reducing the discriminative advantage of content-based detection alone. However, coordina

Same Evidence, Different Answers: Canonical-Context On-Policy Distillation for Multi-Turn Language Models

arXiv:2605.30251v1 Announce Type: cross Abstract: Large language models (LLMs) often solve a task when all instructions are given in a single prompt, but fail when the same information is revealed gra

Securing SIM-Assisted Wireless Networks via Quantum Reinforcement Learning

arXiv:2602.13238v2 Announce Type: replace-cross Abstract: Stacked intelligent metasurfaces (SIMs) have recently emerged as a powerful wave-domain technology that enables multi-stage manipulation of el

Self-Play Reinforcement Learning under Imperfect Information in Big 2

arXiv:2605.28863v1 Announce Type: cross Abstract: Imperfect-information multiplayer games test whether agents can act under hidden information, sparse rewards, and non-stationary opponents. We study t

SGMD: Score Gradient Matching Distillation for Few-Step Video Diffusion Distillation

arXiv:2605.30116v1 Announce Type: new Abstract: Distribution Matching Distillation (DMD) is a widely used paradigm for accelerating inference in few-step video diffusion models. However, DMD-style vid

safetyarxiv-cs-cv

Shame on you Apple, sending @mcuban’s thoughtful email to spam. Seriously? Glad I have learned not to trust your algorithm.

Gary Marcus criticizes Apple's email filtering system for incorrectly routing Mark Cuban's message to spam, highlighting a failure in Apple's spam detection algorithm. The post reflects concerns about

SigmaMedStat: Temporal Signal Modeling for ICU False Alarm Reduction

arXiv:2605.29236v1 Announce Type: new Abstract: Alarm fatigue in intensive care units (ICUs) is a well documented patient safety crisis. Clinical monitors generate 350 or more alarms per patient per d

Statistical Embeddings for Similarity, Retrieval, and Interpretable Alignment of Numeric Tabular Datasets

arXiv:2605.30289v1 Announce Type: new Abstract: Numeric tabular datasets are the dominant data format in scientific practice, yet large language models lack native mechanisms for representing numeric

talking the party line; they switch, he switches. 🤣

This post by cognitive scientist Gary Marcus appears to comment humorously on political or ideological conformity, suggesting that people and leaders flip their positions based on party alignment rath

Teaching Values to Machines: Simulating Human-Like Behavior in LLMs

arXiv:2605.30036v1 Announce Type: new Abstract: Large Language Models (LLMs) demonstrate a remarkable capacity to adopt different personas and roles; however, it remains unclear whether they can manif

The Anatomy of Conversational Scams: A Topic-Based Red Teaming Analysis of Multi-Turn Interactions in LLMs

arXiv:2601.03134v2 Announce Type: replace Abstract: As LLMs gain persuasive capabilities through extended dialogues, they create new opportunities for studying adversarial conversational behavior in e

The Best of the Two Worlds: Harmonizing Semantic and Hash IDs for Sequential Recommendation

arXiv:2512.10388v2 Announce Type: replace-cross Abstract: Conventional Sequential Recommender Systems (SRS) typically assign unique hash IDs (HID) to construct item embeddings, which mainly capture co

The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models

arXiv:2605.29123v1 Announce Type: new Abstract: Masked diffusion language models (MDMs) uniquely support any-order generation, with confidence-based decoding currently serving as the de facto standard

The Importance of Out-of-Band Metadata for Safe Autonomous Agents: The Redpanda Agentic Data Plane

arXiv:2605.29082v1 Announce Type: new Abstract: AI agents are increasingly expected to operate as digital employees: accessing enterprise data, making decisions, and taking actions autonomously. But a

The Sample Complexity of Multiclass and Sparse Contextual Bandits

arXiv:2605.29645v1 Announce Type: cross Abstract: We study contextual bandits in the stochastic i.i.d. setting, where a learner observes contexts drawn from an unknown distribution, selects actions fr

Theoretical Foundations and Effective Algorithms for Policy-Aware Simulator Learning

arXiv:2605.29032v1 Announce Type: new Abstract: Model-based reinforcement learning (MBRL) agents typically learn world models by minimizing predictive loss. However, powerful RL optimizers inevitably

Think Fast, Talk Smart: Partitioning Deterministic and Neural Computation for Structured Health Text Generation

arXiv:2605.29652v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly being used to generate health text from structured records such as wearable time series, biomarkers, vital

this is on track to be this year’s “what did Ilya see?” rumor that people love that has nothing to do with reality.

this is on track to be this year’s “what did Ilya see?” rumor that people love that has nothing to do with reality. Im calling BS on this story. 1. That would be 100,000 employees spending 5k/mo each

Today I learned that @Grimezsz has more courage in her pinky than Roon does in his entire cowardly body. Roon made excuses; Grimes defended …

Today I learned that @Grimezsz has more courage in her pinky than Roon does in his entire cowardly body. Roon made excuses; Grimes defended her views calmly and respectfully, like grownups should. Ope

Tokenmaxxing is so done.

Tokenmaxxing is so done. TOKENS OR HUMANS? The new corporate trade off as AI costs balloon 'This is the first time ever that I can remember that technology costs the same as people' - @jainarvind 'Com

Toward AI Systems That Understand Self and Others: A Multi-Phase Inference Framework for Human Cognitive Diversity and World-Model Alignment

arXiv:2605.29930v1 Announce Type: new Abstract: Mutual misunderstanding in contemporary society does not arise merely because people hold different opinions or values. Even under the same observations

Toward User Preference Alignment in LLM Recommendation via Explicit Context Feedback

arXiv:2605.29141v1 Announce Type: cross Abstract: Traditional recommender systems (RecSys) primarily infer user preferences from implicit signals (such as clicks, watches, and purchases), often neglec

Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

arXiv:2605.29430v1 Announce Type: new Abstract: Automatic speech recognition (ASR) is a core component of human--computer interaction and an increasingly important front-end for LLM-based assistants a

TraceCodec: A Compiler-Backed Neural Codec for Stateful Multi-Flow Network Traffic Traces

arXiv:2605.29941v1 Announce Type: cross Abstract: Critical networking workflows require high-fidelity packet captures (PCAPs) for testing, security analysis, and protocol validation, not just statisti

TRACER: Persistent Regularization for Robust Multimodal Finetuning

arXiv:2605.29380v1 Announce Type: cross Abstract: Mainstream strategies for finetuning pretrained multimodal models often degrade out-of-distribution (OOD) robustness, a phenomenon known as catastroph

Train the Agent, Not the Expert: Learning to Harness Heterogeneous Experts for Multi-Turn Visual Reasoning

arXiv:2605.29894v1 Announce Type: new Abstract: Recent progress in computer vision has produced a wide range of powerful specialized models for detection, segmentation, counting, and other visual task

safetyarxiv-cs-cv

TriSearch: Learning to Optimize Triangulations via Bistellar Flips

arXiv:2605.30220v1 Announce Type: new Abstract: We introduce TriSearch, a reinforcement learning framework for optimizing objectives over triangulations of a polytope via bistellar flips. The key idea

Uncertainty Estimation via Hyperspherical Confidence Mapping

arXiv:2605.05964v2 Announce Type: replace Abstract: Quantifying uncertainty in neural network predictions is essential for high-stakes domains such as autonomous driving, healthcare, and manufacturing

User-Aware Active Knowledge Acquisition for Emotional Support Dialogue

arXiv:2605.29715v1 Announce Type: new Abstract: Emotional support plays an important role in dialogue systems, and its success depends on adapting to a user's evolving and implicit needs across multi-

V2XCrafter: Learning to Generate Driving Scene Across Agents

arXiv:2605.29471v1 Announce Type: new Abstract: Collaborative driving systems leverage vehicle-to-everything (V2X) communication for multi-agent collaborative perception to enhance driving safety, yet

safetyarxiv-cs-cv

ValueFlow: Measuring the Propagation of Value Perturbations in Multi-Agent LLM Systems

arXiv:2602.08567v2 Announce Type: replace-cross Abstract: Multi-agent large language model (LLM) systems increasingly consist of agents that observe and respond to one another's outputs. While value a

VLA-Trace: Diagnosing Vision-Language-Action Models through Representation and Behavior Tracing

arXiv:2605.30117v1 Announce Type: new Abstract: Understanding how Vision-Language-Action (VLA) models transform multimodal knowledge into embodied control remains an open challenge. We present VLA-Tra

Was this priced into the $965 billion valuation?

Was this priced into the $965 billion valuation? This looks like the beginning of the end for OpenAI and Anthropic. The Chinese AI wave did not just cut prices. It destroyed the entire funding logic b

When and How Human Curation Backfires: Preference Alignment under Multi-Model Self-Consuming Loop

arXiv:2605.29267v1 Announce Type: new Abstract: Foundation models are increasingly trained on synthetic data generated by prior model iterations rather than exclusively on real data. This self-consumi

When and How Long? The Readout-Mediator Angle in Temporal Reasoning

arXiv:2605.29126v1 Announce Type: cross Abstract: A linear probe can decode a representation almost perfectly and yet be completely irrelevant to how the model uses it. On calendar-date duration reaso

xModel-KD: Cross-modal Knowledge Distillation for 3D Scene Perception using LiDAR

arXiv:2605.30111v1 Announce Type: cross Abstract: Point cloud segmentation is a fundamental task in 3D scene understanding. Its progress is constrained by the high cost and time required for dense 3D

yes, absolutely, many companies are experimenting. but also: most of those experiments are failing to yield significant RoI. (weird for an e…

yes, absolutely, many companies are experimenting. but also: most of those experiments are failing to yield significant RoI. (weird for an economist to not even ask or address that question.) Looks li

you break it, you buy it peter thiel has broken the united states, and now he is abandoning it

you break it, you buy it peter thiel has broken the united states, and now he is abandoning it Peter Thiel has temporarily relocated his family to Argentina, enrolled his children in school there, and

1. Agreed that OpenAI is in deep trouble; that’s why I have long suggested that it might be the WeWork of AI but 2. Anthropic is not out of …

1. Agreed that OpenAI is in deep trouble; that’s why I have long suggested that it might be the WeWork of AI but 2. Anthropic is not out of the woods; their best quarter was exactly when tokenmaxxing

A Comparative Study of Rule-Based and Data-Driven Approaches in Industrial Monitoring

arXiv:2509.15848v2 Announce Type: replace Abstract: Industrial monitoring systems, especially when deployed in Industry 4.0 environments, are experiencing a shift in paradigm from traditional rule-bas

A Factory-Floor Deployment Case Study of VLA Pipelines for Industrial Packaging Task: Workflow, Failures, and Lessons

arXiv:2605.27461v1 Announce Type: new Abstract: Vision-Language-Action (VLA) policies have shown promising manipulation capabilities, yet their practical impact is often limited by the reliability dem

safetyarxiv-cs-ro

A Policy-Driven Runtime Layer for Agentic LLM Serving

arXiv:2605.27744v1 Announce Type: new Abstract: Multi-agent LLM systems have become the dominant production workload, but the serving stack was not built for them. The agent framework above knows agen

A Structural Theory of Position Bias in Transformers

arXiv:2602.16837v2 Announce Type: replace Abstract: Transformer models systematically favor certain token positions, yet the architectural origins of this position bias remain poorly understood. This

Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection

arXiv:2605.28664v1 Announce Type: cross Abstract: Safety detection models require examples of HHH (Helpful, Harmless, Honest)-violating outputs for robust generalization, however such examples are sca