Search: “safety”

Grok just hit its highest monthly traffic EVER: over 326 MILLION visits in March alone That’s a massive 61% jump YoY and up 9.3% just since …

Grok just hit its highest monthly traffic EVER: over 326 MILLION visits in March alone That’s a massive 61% jump YoY and up 9.3% just since February People love Grok because it's the only AI they can

safetyelon-musk--x

Safety

GUIDE: Guided Updates for In-context Decision Evolution in LLM-Driven Spacecraft Operations

DGX agent

arXiv:2603.27306v2 Announce Type: replace-cross Abstract: Large language models (LLMs) have been proposed as supervisory agents for spacecraft operations, but existing approaches rely on static prompt

safetyarxiv-cs-ai

15 Apr 2026

Hail to the Thief: Exploring Attacks and Defenses in Decentralised GRPO

arXiv:2511.09780v2 Announce Type: replace Abstract: Group Relative Policy Optimization (GRPO) has demonstrated wide adoption in the post-training of Large Language Models (LLMs). In GRPO, prompts are

HiCoLoRA: Addressing Context-Prompt Misalignment via Hierarchical Collaborative LoRA for Zero-Shot DST

arXiv:2509.19742v4 Announce Type: replace-cross Abstract: Zero-shot Dialog State Tracking (zs-DST) is essential for enabling Task-Oriented Dialog Systems (TODs) to generalize to new domains without co

Human-Centric Topic Modeling with Goal-Prompted Contrastive Learning and Optimal Transport

arXiv:2604.12663v1 Announce Type: new Abstract: Existing topic modeling methods, from LDA to recent neural and LLM-based approaches, which focus mainly on statistical coherence, often produce redundan

Hybrid-AIRL: Enhancing Inverse Reinforcement Learning with Supervised Expert Guidance

arXiv:2511.21356v2 Announce Type: replace-cross Abstract: Adversarial Inverse Reinforcement Learning (AIRL) has shown promise in addressing the sparse reward problem in reinforcement learning (RL) by

I find it fascinating that OpenAI's Global Affairs team will denounce AI doomers for being too negative and then, simultaneously, advocate f…

I find it fascinating that OpenAI's Global Affairs team will denounce AI doomers for being too negative and then, simultaneously, advocate for an Illinois bill that would shield them from liability if

safetygary-marcus--x

I spent some time trying to distill all the complex factors impacting open models -- economics, capabilities, distribution, policy, etc. -- …

I spent some time trying to distill all the complex factors impacting open models -- economics, capabilities, distribution, policy, etc. -- into a clear list of beliefs. Here they are in full. 1. It’s

safetyyann-lecun--x

Incentivizing High-Quality Human Annotations with Golden Questions

arXiv:2505.19134v2 Announce Type: replace-cross Abstract: Human-annotated data plays a vital role in training large language models (LLMs), such as supervised fine-tuning and human preference alignmen

Information-Geometric Decomposition of Generalization Error in Unsupervised Learning

arXiv:2604.12340v1 Announce Type: cross Abstract: We decompose the Kullback--Leibler generalization error (GE) -- the expected KL divergence from the data distribution to the trained model -- of unsup

InsightFlow: LLM-Driven Synthesis of Patient Narratives for Mental Health into Causal Models

arXiv:2604.12721v1 Announce Type: new Abstract: Clinical case formulation organizes patient symptoms and psychosocial factors into causal models, often using the 5P framework. However, constructing su

Labeled TrustSet Guided: Batch Active Learning with Reinforcement Learning

arXiv:2604.12303v1 Announce Type: new Abstract: Batch active learning (BAL) is a crucial technique for reducing labeling costs and improving data efficiency in training large-scale deep learning model

Learning step-level dynamic soaring in shear flow

arXiv:2604.12413v1 Announce Type: cross Abstract: Dynamic soaring enables sustained flight by extracting energy from wind shear, yet it is commonly understood as a cycle-level maneuver that assumes st

Learning Versatile Humanoid Manipulation with Touch Dreaming

arXiv:2604.13015v1 Announce Type: new Abstract: Humanoid robots promise general-purpose assistance, yet real-world humanoid loco-manipulation remains challenging because it requires whole-body stabili

Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation

arXiv:2604.13010v1 Announce Type: cross Abstract: On-policy distillation (OPD) has emerged as an efficient post-training paradigm for large language models. However, standard OPD requires a live teach

LiveMoments: Reselected Key Photo Restoration in Live Photos via Reference-guided Diffusion

arXiv:2604.12286v1 Announce Type: new Abstract: Live Photo captures both a high-quality key photo and a short video clip to preserve the precious dynamics around the captured moment. While users may c

Man and machine: artificial intelligence and judicial decision making

arXiv:2603.19042v4 Announce Type: replace Abstract: The integration of artificial intelligence (AI) technologies into judicial decision-making, particularly in pretrial, sentencing, and parole context

Meet Dynamic Individual Preferences: Resolving Conflicting Human Value with Paired Fine-Tuning

arXiv:2604.12479v1 Announce Type: new Abstract: Recent advances in large language models (LLMs) have significantly improved the alignment of models with general human preferences. However, a major cha

Models Know Their Shortcuts: Deployment-Time Shortcut Mitigation

arXiv:2604.12277v1 Announce Type: new Abstract: Pretrained language models often rely on superficial features that appear predictive during training yet fail to generalize at test time, a phenomenon k

MODIX: A Training-Free Multimodal Information-Driven Positional Index Scaling for Vision-Language Models

arXiv:2604.12537v1 Announce Type: cross Abstract: Vision-Language Models (VLMs) have achieved remarkable progress in multimodal understanding, yet their positional encoding mechanisms remain suboptima

MolMem: Memory-Augmented Agentic Reinforcement Learning for Sample-Efficient Molecular Optimization

arXiv:2604.12237v1 Announce Type: cross Abstract: In drug discovery, molecular optimization aims to iteratively refine a lead compound to improve molecular properties while preserving structural simil

Mutual Information Surprise: Rethinking Unexpectedness in Autonomous Systems

arXiv:2508.17403v3 Announce Type: replace Abstract: A community of researchers appears to think that a machine can be surprised and have introduced various surprise measures, principally the Shannon S

No More Stale Feedback: Co-Evolving Critics for Open-World Agent Learning

arXiv:2601.06794v2 Announce Type: replace Abstract: Critique-guided reinforcement learning (RL) has emerged as a powerful paradigm for training LLM agents by augmenting sparse outcome rewards with nat

Not All Turns Are Equally Hard: Adaptive Thinking Budgets For Efficient Multi-Turn Reasoning

arXiv:2604.05164v2 Announce Type: replace-cross Abstract: As LLM reasoning performance plateau, improving inference-time compute efficiency is crucial to mitigate overthinking and long thinking traces

Offline-Online Reinforcement Learning for Linear Mixture MDPs

arXiv:2604.11994v1 Announce Type: new Abstract: We study offline-online reinforcement learning in linear mixture Markov decision processes (MDPs) under environment shift. In the offline phase, data ar

PAINT: Partner-Agnostic Intent-Aware Cooperative Transport with Legged Robots

arXiv:2604.12852v1 Announce Type: new Abstract: Collaborative transport requires robots to infer partner intent through physical interaction while maintaining stable loco-manipulation. This becomes pa

Perception-Aware Policy Optimization for Multimodal Reasoning

arXiv:2507.06448v5 Announce Type: replace Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has proven to be a highly effective strategy for endowing Large Language Models (LLMs) with ro

PR-MaGIC: Prompt Refinement Via Mask Decoder Gradient Flow For In-Context Segmentation

arXiv:2604.12113v1 Announce Type: cross Abstract: Visual Foundation Models (VFMs) such as the Segment Anything Model (SAM) have significantly advanced broad use of image segmentation. However, SAM and

Prediction from a year ago about AI backlash that looks to be on track:

Prediction from a year ago about AI backlash that looks to be on track: Public backlash against AI will so be strong by 2028 that anti-AI sentiment will probably be a major factor in the 2028 US Presi

safetygary-marcus--x

Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation

arXiv:2511.17097v2 Announce Type: replace Abstract: Vision-Language Navigation requires agents to act coherently over long horizons by understanding not only local visual context but also how far they

PubSwap: Public-Data Off-Policy Coordination for Federated RLVR

arXiv:2604.12160v1 Announce Type: new Abstract: Reasoning post-training with reinforcement learning from verifiable rewards (RLVR) is typically studied in centralized settings, yet many realistic appl

Redefining Quality Criteria and Distance-Aware Score Modeling for Image Editing Assessment

arXiv:2604.12175v1 Announce Type: new Abstract: Recent advances in image editing have heightened the need for reliable Image Editing Quality Assessment (IEQA). Unlike traditional methods, IEQA require

Relaxing Anchor-Frame Dominance for Mitigating Hallucinations in Video Large Language Models

arXiv:2604.12582v1 Announce Type: new Abstract: Recent Video Large Language Models (Video-LLMs) have demonstrated strong capability in video understanding, yet they still suffer from hallucinations. E

Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe

arXiv:2604.13016v1 Announce Type: cross Abstract: On-policy distillation (OPD) has become a core technique in the post-training of large language models, yet its training dynamics remain poorly unders

Retrieval as a Decision: Training-Free Adaptive Gating for Efficient RAG

arXiv:2511.09803v2 Announce Type: replace Abstract: Retrieval-Augmented Generation (RAG) improves factuality but retrieving for every query often hurts quality while inflating tokens and latency. We p

SAM3-I: Segment Anything with Instructions

arXiv:2512.04585v3 Announce Type: replace Abstract: Segment Anything Model 3 (SAM3) advances open-vocabulary segmentation through promptable concept segmentation, enabling users to segment all instanc

Scaffold-Conditioned Preference Triplets for Controllable Molecular Optimization with Large Language Models

arXiv:2604.12350v1 Announce Type: cross Abstract: Molecular property optimization is central to drug discovery, yet many deep learning methods rely on black-box scoring and offer limited control over

Scalable and General Whole-Body Control for Cross-Humanoid Locomotion

arXiv:2602.05791v2 Announce Type: replace Abstract: Learning-based whole-body controllers have become a key driver for humanoid robots, yet most existing approaches require robot-specific training. In

Schema-Adaptive Tabular Representation Learning with LLMs for Generalizable Multimodal Clinical Reasoning

arXiv:2604.11835v1 Announce Type: cross Abstract: Machine learning for tabular data remains constrained by poor schema generalization, a challenge rooted in the lack of semantic understanding of struc

Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision

arXiv:2604.12002v1 Announce Type: new Abstract: Current post-training methods in verifiable settings fall into two categories. Reinforcement learning (RLVR) relies on binary rewards, which are broadly

Simulation as Supervision: Mechanistic Pretraining for Scientific Discovery

arXiv:2507.08977v4 Announce Type: replace-cross Abstract: Scientific modeling faces a tradeoff between the interpretability of mechanistic theory and the predictive power of machine learning. While ex

So true. “Being right too soon is socially unacceptable”

So true. “Being right too soon is socially unacceptable” I remember reading this from Heinlein as a young man and it took experience for me to fully understand it. People and institutions will persist

safetygary-marcus--x

SOAR: Self-Correction for Optimal Alignment and Refinement in Diffusion Models

arXiv:2604.12617v1 Announce Type: cross Abstract: The post-training pipeline for diffusion models currently has two stages: supervised fine-tuning (SFT) on curated data and reinforcement learning (RL)

StableSketcher: Enhancing Diffusion Model for Pixel-based Sketch Generation via Visual Question Answering Feedback

arXiv:2510.20093v2 Announce Type: replace-cross Abstract: Although recent advancements in diffusion models have significantly enriched the quality of generated images, challenges remain in synthesizin

Task Alignment: A simple and effective proxy for model merging in computer vision

arXiv:2604.12935v1 Announce Type: new Abstract: Efficiently merging several models fine-tuned for different tasks, but stemming from the same pretrained base model, is of great practical interest. Des

Teaching LLMs Human-Like Editing of Inappropriate Argumentation via Reinforcement Learning

arXiv:2604.12770v1 Announce Type: new Abstract: Editing human-written text has become a standard use case of large language models (LLMs), for example, to make one's arguments more appropriate for a d

The front door to the internet hasn't changed -- but the person going through that door has changed from a human browsing 5 blue links, to a…

The front door to the internet hasn't changed -- but the person going through that door has changed from a human browsing 5 blue links, to an agent browsing a similar index in vastly different ways. A

safetysonya-huang--x

The role of System 1 and System 2 semantic memory structure in human and LLM biases

arXiv:2604.12816v1 Announce Type: new Abstract: Implicit biases in both humans and large language models (LLMs) pose significant societal risks. Dual process theories propose that biases arise primari