Search: “safety”

Crazy to see this. I have spent years saying that GenAI was unreliable and we can’t trust it. The fact that it cannot be trusted to follow i…

Crazy to see this. I have spent years saying that GenAI was unreliable and we can’t trust it. The fact that it cannot be trusted to follow instructions is a core part of the issue, as I have been poin

safetygary-marcus--x

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

CROCODIL: Cross-Model Code Editing with LLMs

DGX agent

arXiv:2609.03894v1 Announce Type: new Abstract: Large language models (LLMs) have become ubiquitous tools for code generation and editing. However, development teams often use multiple LLM assistants.

safetyarxiv-cs-cl

4 Sept 2026

DE-Venus: A Data-Efficient RLVR Framework for Large Language Models

arXiv:2609.03324v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) improves large language model reasoning, but its practical scaling is constrained by expensive on-

Decoupled Analysis-Judging: An Automated Creativity Evaluator Using LLMs in Complex Multi-step Creativity Tasks

arXiv:2609.03432v1 Announce Type: new Abstract: Automated evaluation of creativity tasks remains challenging for LLM-as-a-Judge, as LLM is susceptible to biases such as verbosity bias and leniency bia

Differentiable Hybrid Modelling for Learning and Optimising Chemical Transport Processes from Experimental Data

arXiv:2609.04011v1 Announce Type: cross Abstract: Reliable transport models are essential when modelling and optimising many chemical engineering processes, yet, most models assume hand-picked constit

Differentiable Interval Bottlenecks for Interpretable Anomaly Detection in Numerical Data

arXiv:2609.03878v1 Announce Type: cross Abstract: Reconstruction-based anomaly detectors are accurate but opaque: a deep autoencoder flags a sample without telling a practitioner which feature ranges

Does Astra have human-like continual learning capability over long periods? If not, it's not AGI. Even a system that could pass @demishassab…

Does Astra have human-like continual learning capability over long periods? If not, it's not AGI. Even a system that could pass @demishassabis's new 'Turing test' where it has to independently discove

safetygary-marcus--x

DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training

arXiv:2609.04094v1 Announce Type: new Abstract: Reinforcement Learning from Verifiable Rewards works well when a task has a programmatic checker, but most long-horizon agent domains have none. We work

EF1-Constrained Nash Social Welfare with Identical Additive Valuations: Complexity, Guarantees, and Experiments

arXiv:2609.03846v1 Announce Type: cross Abstract: We study the allocation of indivisible goods among agents with identical additive valuations, focusing on envy-freeness up to one good (EF1) and Nash

Exploring the Potential of Contrastive Language-Image Pre-training for Multi-Source Remote Sensing Data

arXiv:2609.03391v1 Announce Type: cross Abstract: Contrastive language-image learning (CLIP) has become a key paradigm for remote sensing vision-language understanding. However, existing remote sensin

FlashRender: Few-Step Generative Rendering via Camera-Controlled Video MeanFlow

arXiv:2609.03563v1 Announce Type: new Abstract: We present FlashRender, a few-step generative rendering framework that retakes a source video along a target camera trajectory in seconds. We identify s

FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience

arXiv:2609.03241v1 Announce Type: cross Abstract: A reasoning model can improve from its own on-policy experience, but this inner loop is fragile: terminal verifiers provide reliable yet sparse superv

From Prior-Guided Heuristics to Deployable Agents: Accelerating Demonstration-Driven Reinforcement Learning for Deadline-Constrained Network Control

arXiv:2609.03590v1 Announce Type: cross Abstract: Timely delivery of delay-sensitive information over dynamic, heterogeneous networks is essential for NextG interactive applications, yet providing str

FWBC-VLA: Force-Aware Whole-Body Compensation for Contact-Rich Loco-Manipulation

arXiv:2609.03889v1 Announce Type: cross Abstract: Contact-rich loco-manipulation requires a bridge between semantic action generation and physical interaction control. Existing Vision-language-action

GazeFS: Target-Centered Gaze-Trajectory Forecasting and Stabilization from Gaze-Head History

arXiv:2609.03868v1 Announce Type: cross Abstract: Target-centered gaze interaction requires more than suppressing frame-to-frame fluctuations: target acquisition produces task-aligned changes in gaze-

GIFT: Guided Intermediate Feature Training via Action-Oriented Structural Supervision for Robotic Manipulation

arXiv:2609.04193v1 Announce Type: new Abstract: Vision-language pre-training and predictive world modeling provide robot policies with rich semantic and dynamic visual features, but their native actio

Good old days at OpenAI in 2016: an agent stares at screen pixels, moves a mouse, and books a flight on United. We called it World of Bits, …

Good old days at OpenAI in 2016: an agent stares at screen pixels, moves a mouse, and books a flight on United. We called it World of Bits, inside OpenAI Universe. 10 yrs later, Astra is reincarnated

safetyjim-fan--x

Gradients Know What Outcomes Don't: Unlocking Reinforcement Learning for LLM Reasoning with Gradient-Aligned Rewards

arXiv:2609.03342v1 Announce Type: new Abstract: Reinforcement learning from verifiable rewards (RLVR) drives chain-of-thought reasoning in large language models, yet its binary outcome reward cannot d

Hadronic Mono-Z Dark Matter Sensitivity with Flow Matching on CMS Open Data

arXiv:2609.02923v1 Announce Type: cross Abstract: We present a projected sensitivity study for hadronic mono-Z dark-matter production using CMS Run~2015D HTMHT open data corresponding to 2.256382381~i

Headroom-Drift Replay: A Primitive for Principled Replay Control in GRPO

arXiv:2609.03941v1 Announce Type: cross Abstract: RL-based post-training for reasoning models is increasingly bottlenecked by repeated fresh rollout generation, particularly in agentic settings where

High-Altitude Balloon Station-Keeping with First Order Model Predictive Control

arXiv:2511.07761v2 Announce Type: replace Abstract: High-altitude balloons (HABs) are common in scientific research due to their wide range of applications and low cost. Because of their nonlinear, un

High-Dimensional Learning Dynamics of Attention-Indexed Models

arXiv:2609.03858v1 Announce Type: new Abstract: Attention mechanisms are central to modern foundation models, yet their training dynamics remain poorly understood, especially when the attention matric

How Intuit built an agentic disaster recovery assistant with Amazon Bedrock

Disaster recovery at scale is hard. Learn how Intuit built EWOK Agent, an agentic disaster recovery assistant on Amazon Bedrock that lets on-call engineers run production failovers from a plain-langua

safetyaws-ml-blog

IchthyoNoma: Nomenclature and Context Sensitivity of Zero-Shot Biological Vision--Language Models for Bangladeshi Freshwater Fish Recognition

arXiv:2609.03985v1 Announce Type: cross Abstract: Zero-shot vision-language models (VLMs) are increasingly used as training-free species recognizers, but reported accuracy can reflect more than visual

Imagine-then-Plan: Agent Learning from Adaptive Lookahead with World Models

arXiv:2601.08955v3 Announce Type: replace-cross Abstract: Recent advances in world models have shown promise for modeling future dynamics of environmental states, enabling agents to reason and act wit

InSituMeasure: Probing Situated Measurement Grounding in Industrial Scenes with Multimodal Large Language Models

arXiv:2609.04014v1 Announce Type: new Abstract: For trained operators, gauge reading requires little specialized knowledge, low cognitive effort, and high repeatability. Yet Multimodal Large Language

It's weird thinking back to 6 weeks ago, when I and others thanked OpenAI for disclosing the HF incident, now that we know they were coverin…

It's weird thinking back to 6 weeks ago, when I and others thanked OpenAI for disclosing the HF incident, now that we know they were covering up a 2nd incident. Are there more incidents that they know

safetygary-marcus--x

Large Language Models in Resolving Contextual Knowledge Conflicts

arXiv:2609.03148v1 Announce Type: new Abstract: Most prior works focused on conflicts between an LLM's internal parametric knowledge and externally provided context. In contrast, we investigate how LL

LeanGRPO: Eliminating Redundant Recomputation in Diffusion RL

arXiv:2609.03528v1 Announce Type: cross Abstract: Diffusion reinforcement learning (RL) has recently achieved significant success in post-training image and video generative models. However, most diff

Less Is Moral: A CHARMing Framework for Moral Foundations Detection in Endorsement Behaviour

arXiv:2609.03330v1 Announce Type: new Abstract: Moral language plays a central role in shaping online endorsement and the diffusion of information, yet existing moral foundation detection systems ofte

LLM as GNN: Graph Vocabulary Learning for Text-Attributed Graph Foundation Models

arXiv:2503.03313v4 Announce Type: replace-cross Abstract: Text-Attributed Graphs (TAGs), where each node is associated with text descriptions, are ubiquitous in real-world scenarios. They typically ex

MasterControl Seventeen Every Time

arXiv:2609.03209v1 Announce Type: new Abstract: We study a governed approach to enterprise analytics: a language model interprets the question, while deterministic policy selects and runs a pre-approv

Mudragen: Geometrically Supervised Generation of Interacting Two-Hand Mudras for Preserving Indian Classical Dance Heritage

arXiv:2609.03415v1 Announce Type: new Abstract: Automatic generation of hand gestures is essential for the transmission of Indian classical dance and critical for its preservation. Indian classical da

MulDP: Multimodal Diffusion Policy for Autonomous Quadruped Parkour Navigation across Complex Terrains

arXiv:2609.03984v1 Announce Type: new Abstract: Quadruped robots have demonstrated impressive agility in parkour locomotion across complex terrains. However, most systems still rely on human intervent

Multi-step Proximal Policy Improvement in Offline Reinforcement Learning

arXiv:2609.03842v1 Announce Type: new Abstract: Offline reinforcement learning (RL) must reconcile two competing requirements: policy updates should stay near dataset-supported actions to keep value e

NeoRed: A Knowledge-Logic-Alignment Multimodal Large Language Model for Neonatal Respiratory Disease Diagnosis

arXiv:2609.03527v1 Announce Type: new Abstract: Neonatal respiratory diseases are a major cause of neonatal morbidity and mortality, posing substantial challenges in clinical practice. Despite recent

Neural Video Compression Based on Deformable Temporal Alignment and Difference-aware Fusion

arXiv:2609.03520v1 Announce Type: cross Abstract: In conditional coding-based neural video compression, the quality of temporal context directly affects compression per- formance. Existing methods mos

No country for old linguists: LLM-brain alignment underdetermines neural computation

arXiv:2609.03160v1 Announce Type: new Abstract: Nastase et al. (2026) argue that large language models (LLMs) may illuminate language processing because both rely on distributed, context-sensitive rep

Not All Preferences Deserve Gradients: Understanding Gradient Utility in Offline Reasoning Alignment

arXiv:2602.01207v2 Announce Type: replace Abstract: Offline preference optimization aligns reasoning models from fixed chosen--rejected pairs, yet standard methods apply gradient updates from every pa

PaperScout: An Autonomous Agent for Academic Paper Search with Process-Aware Sequence-Level Policy Optimization

arXiv:2601.10029v3 Announce Type: replace Abstract: Academic paper search is a fundamental task in scientific research, yet most existing approaches organize retrieval around predefined workflows or s

Portable Causal Fairness Across Synthetic Data Generator Families

arXiv:2609.03180v1 Announce Type: new Abstract: When a statistical agency or regulator releases synthetic data in place of sensitive records, it chooses the generator that produces the table, and can

PoseDreamer: Scalable and Photorealistic Human Data Generation Pipeline with Diffusion Models

arXiv:2603.28763v2 Announce Type: replace Abstract: Acquiring labeled datasets for 3D human mesh estimation is challenging due to depth ambiguities and the inherent difficulty of annotating 3D geometr

PPO-STGNN: A Proximal Policy Optimization Approach with Spatio-Temporal Graph Neural Networks for DAG Task Scheduling in Cloud-Edge-End Computing

arXiv:2609.03503v1 Announce Type: new Abstract: With the rapid development of the Internet of Things, computation intensive directed acyclic graph (DAG) tasks have become increasingly common in cloud-

Privacy, Robustness, and Fairness Trade-offs in Federated Intrusion Detection: Geometric Indistinguishability at the Aggregation Interface

arXiv:2609.03420v1 Announce Type: cross Abstract: Federated learning enables privacy-conscious collaboration for network intrusion detection without centralizing sensitive traffic data, yet its deploy

Quantum Implicit Neural Representations for Novel View Synthesis

arXiv:2601.05250v2 Announce Type: replace Abstract: Quantum implicit neural representations have recently emerged as compact function approximators for continuous visual signals. In parallel, Neural R

R2S-Eval: Robot Evaluation with Real-to-Sim Calibration via Vision-Language Models

arXiv:2609.03276v1 Announce Type: new Abstract: Evaluating robot manipulation policies is becoming increasingly important as generalist models, particularly vision-language-action (VLA) models, are de

Remember and Reweight: Enhancing Multi-Agent Debate with Experience Memory and Confidence Estimation

arXiv:2609.03619v1 Announce Type: cross Abstract: Multi-agent debate (MAD) improves the reasoning capabilities of large language models by having multiple agents iteratively refine their responses thr

Rethinking On-Policy Distillation of Large Language Models II: One Training Example

arXiv:2609.04172v1 Announce Type: new Abstract: On-policy distillation (OPD) combines student-generated rollouts with dense token-level supervision from a teacher. Existing work has mainly studied its