safety

Towards Understanding the Training of GFlowNets: A Theoretical Perspective

arXiv:2610.13711v1 Announce Type: new Abstract: Generative Flow Networks (GFlowNets) learn policies that sample discrete objects in proportion to a positive reward. They are trained by minimizing bala

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Tracing and Taming Hallucinated Terms via Subsequence-Association Filtering and Breadth-Aware Selection for LLM Query Expansion

DGX agent

arXiv:2610.13810v1 Announce Type: cross Abstract: Large language models (LLMs) are now a standard source of query expansion terms but they hallucinate. They inject non existent entities and off-topic

safetyarxiv-cs-cl

12 Oct 2026

Tracking the Invisible: Self-Supervised Wheel Track Detection with Thermal-RGB Fusion

arXiv:2610.13872v1 Announce Type: cross Abstract: Accurate detection of drivable area is essential for autonomous driving and advanced driver assistance systems. While state-of-the-art methods focus o

UniConFlow: A Unified Constrained Flow-Matching Framework for Certified Motion Planning

arXiv:2506.02955v3 Announce Type: replace-cross Abstract: Generative models have become increasingly powerful tools for robot motion generation, enabling flexible and multimodal trajectory generation

Universal Bounds for Self-Normalized Kullback-Leibler Estimation

arXiv:2610.12851v1 Announce Type: cross Abstract: This work studies an entropy-based estimator of Kullback-Leibler (KL) divergence constructed from normalized importance weights. We prove a universal

Variance-Reduced Policy Gradient is Nearly Minimax-Optimal

arXiv:2610.13692v1 Announce Type: cross Abstract: Standard Policy Gradient (PG) algorithms suffer from high estimation variance because they rely on Monte Carlo updates. While off-policy variance redu

VEPA: Verified Evolving Formal Policy for Lifelong LLM Agent Guardrails

arXiv:2610.12813v1 Announce Type: new Abstract: LLM agents increasingly operate over sensitive resources, such as private documents and mutable system states, which can create serious safety and secur

Vision Transformers Learn Word Meaning from Images Alone

arXiv:2610.13252v1 Announce Type: cross Abstract: Can a vision model learn what words mean from pixels alone? We show that the two largest DINOv3 vision transformers organize words rendered in images

Visual-Wi-Fi Navigation to Signal-Emitting Devices

arXiv:2610.12733v1 Announce Type: new Abstract: Indoor navigation to task-relevant targets remains difficult when the target is small, occluded, or visually indistinct. Meanwhile, smart appliances, mo

VLMs are Native Action Models: Unleashing the Manipulation Capabilities of VLMs with Explicit Action-Space Grounding

arXiv:2610.12832v1 Announce Type: new Abstract: Pretrained vision-language models (VLMs) already exhibit strong visual, spatial and reasoning capabilities relevant to robotic manipulation. The key cha

We need a name for the kind of people that conflate 2022 pure LLMs that were in fact a tiny bit like stochastic parrots (not my phrase, and …

We need a name for the kind of people that conflate 2022 pure LLMs that were in fact a tiny bit like stochastic parrots (not my phrase, and not one I ever endorsed) with 2026 systems that couple those

safetygary-marcus--x

What Do Audio-Language Judges of Generated Music Listen For? Four Probes with Controls from Expert Ratings

arXiv:2610.13247v1 Announce Type: cross Abstract: Open audio-language models are used as judges of generated music: given a clip and its text prompt, the model rates the clip, and the judge is validat

What Should SFT Learn? Conditional Information Gain for Generalizable Post-Training

arXiv:2610.12995v1 Announce Type: cross Abstract: Supervised fine-tuning (SFT) remains a widely used approach for post-training large language models, because fixed demonstrations can directly introdu

When MLLMs miss what humans read: diagnosing and learning to read squint text

arXiv:2610.13465v1 Announce Type: cross Abstract: Multimodal LLMs can fail to read text embedded in an image's coarse structure (text a human recovers by squinting) and report that no text is present.

Why Linear Models Can Still Work in a Complex and Changing World

arXiv:2610.13404v1 Announce Type: new Abstract: Linear models can predict well in complex domains even when underlying mechanisms are nonlinear. We argue that this can occur because the complexity of

X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation

arXiv:2610.13687v1 Announce Type: cross Abstract: Audio-visual dataset distillation aims to synthesize a small dataset from a large paired audio-visual dataset, while nearly preserving the training ut

Your Discrete Graph Diffusion Model is Secretly Memorizing

arXiv:2610.13786v1 Announce Type: cross Abstract: Discrete graph diffusion models (GDMs) corrupt the graph structure by adding noise independently to nodes and edges under a Markovian noise model. In

Introducing Multi-Agent Self-Supervision (MASS): Recursive self-improvement through collective intelligence Blog: https://pub.sakana.ai/mass…

Introducing Multi-Agent Self-Supervision (MASS): Recursive self-improvement through collective intelligence Blog: https://pub.sakana.ai/mass/ As AI systems tackle increasingly open-ended research prob

safetydavid-ha--x

11 Oct 2026

A look at a 1,700-member Slack run by Medicare agency CMS where Microsoft, OpenAI, and other companies help shape policy on AI apps and medical records access (CBS News)

CBS News: A look at a 1,700-member Slack run by Medicare agency CMS where Microsoft, OpenAI, and other companies help shape policy on AI apps and medical records access — A cadre of elite techies, fin

safetytechmeme

10 Oct 2026

Reasoning From Scratch: Reinforcement Learning with Verifiable Rewards (RLVR) round 2. Covering clipped policy ratios, KL loss term, format …

safetysebastian-raschka--x

Reasoning From Scratch: Reinforcement Learning with Verifiable Rewards (RLVR) round 2. Covering clipped policy ratios, KL loss term, format rewards, and other GRPO tips & tricks. 00:00 Introduction an

10 Oct 2026

A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control

arXiv:2610.12465v1 Announce Type: cross Abstract: General-purpose robots must perform a wide range of tasks from agile locomotion to dexterous manipulation. While sim-to-real reinforcement learning (R

A Geometric Approach to Soft Actor-Critic with Zonotopes for Locomotion Learning

arXiv:2610.12113v1 Announce Type: cross Abstract: Off-policy actor--critic methods control overestimation bias by taking the minimum of two critics. This uses the same aggregation rule everywhere, reg

A Physics-Informed Collision Learning Framework for Collaborative Robot Motion Generation

arXiv:2610.12404v1 Announce Type: new Abstract: Close-proximity multi-arm manipulation requires collision models that are both geometrically accurate and differentiable enough for real-time optimizati

A Unified Bellman Operator for Safety-Critical Reinforcement Learning

arXiv:2610.12420v1 Announce Type: new Abstract: Reinforcement learning in safety-critical domains requires maximizing task performance while strictly adhering to safety constraints. Existing safe rein

ActiveMedAgent: Cost-Aware Trajectory Learning for Multimodal Medical Diagnosis

arXiv:2610.11140v1 Announce Type: cross Abstract: Clinical diagnosis is inherently sequential: clinicians escalate from cheap to costly tests only when additional evidence is expected to resolve diagn

AdaptEvo: Adaptive Agent Learning with Evolving Supervision

arXiv:2610.11354v1 Announce Type: new Abstract: Rule-governed contextual decision tasks require models to apply specified rules to case-specific context and evidence. Written rules can leave gaps in d

Addressing Overcommitment in the Reasoning of Gendered Economic Memes under Multimodal Ambiguity

arXiv:2610.11724v1 Announce Type: new Abstract: Multimodal meme understanding is increasingly used to analyze socially sensitive content, yet existing models often exhibit biased behavior when interpr

AffordDrive3D: Affordance-Aware World-Action Modeling with Spatial Understanding

arXiv:2610.11060v1 Announce Type: cross Abstract: World-action models have recently improved autonomous driving by jointly learning future scene prediction and trajectory generation. Most existing app

Agentic Critical Training

arXiv:2603.08706v2 Announce Type: replace Abstract: Imitation learning (IL) teaches language-model agents to reproduce expert actions but not to distinguish them from plausible mistakes. Self-reflecti

Alignment is a human-AI workflow: how Custom Views make agent review possible

Automated evals can score every agent session, but they do not automatically reflect what your team considers a good outcome. That takes human judgment. The post Alignment is a human-AI workflow: how

safetyarize-ai

AliO: Output Alignment Matters in Long-Term Time Series Forecasing

arXiv:2610.11213v1 Announce Type: new Abstract: Long-term Time Series Forecasting (LTSF) tasks, which leverage the current data sequence as input to predict the future sequence, have become increasing

All Verdicts are Not Equal: Rethinking LLM Judge Reliability

arXiv:2610.12083v1 Announce Type: cross Abstract: LLM-as-a-Judge is the standard paradigm for NLP evaluation, yet its systemic reliability remains poorly understood despite being widely treated as a d

Anthropic is setting up a 'presidential engagement' program for the 2028 US elections that will offer AI policy education to candidates in both parties (Emily Forlini/Fortune)

Emily Forlini / Fortune: Anthropic is setting up a “presidential engagement” program for the 2028 US elections that will offer AI policy education to candidates in both parties — Anthropic is setting

safetytechmeme

ARC: A Reasoning Recipe for Robot Foundation Models

arXiv:2610.12386v1 Announce Type: cross Abstract: The prevailing approach to improving robot foundation models (RFMs) relies on larger models, more robot demonstrations, and costly training at scale.

Assorted things I've been mulling over regarding agents in the economy: • Companies are in some sense the original superintelligence. It's o…

Assorted things I've been mulling over regarding agents in the economy: • Companies are in some sense the original superintelligence. It's often the case that they can use their amortized cognitive su

safetyelon-musk--x

AuraLuxMuse: Adaptive Fusion Modeling for Aesthetic Stage Lighting Design with Music and Expert Guidance

arXiv:2610.11792v1 Announce Type: cross Abstract: We present AuraLuxMuse, a novel system for automated aesthetic stage lighting design that integrates expert knowledge, representation learning, and pr

Balancing Reference Guidance and Free Generation in Trajectory Rollouts for Reasoning RL

arXiv:2610.11128v1 Announce Type: new Abstract: A verified reference solution provides a correct trajectory for training a reasoning model. Alternatively, a prefix of the reference can guide the model

Beyond Euclidean Clipping: Overcoming Exploration Collapse in LLM RL via Riemannian Isometric Policy Optimization

arXiv:2607.10169v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become a dominant paradigm for enhancing LLMs' reasoning capabilities. However, RL algorithms with PPO-Clip ar

Beyond Speech Captions: Speech-Rewarded Style Planning for Conversational Text-to-Speech

arXiv:2610.11461v1 Announce Type: cross Abstract: Natural-language style descriptions provide an interpretable interface between large language models (LLMs) and controllable text-to-speech (TTS). How

Beyond the Ergodic Wall: A Discrete Geometric Physics Sandbox for Analysing AI Scaling Limits and Complexity Collapse

arXiv:2610.10651v1 Announce Type: cross Abstract: This paper exposes the ergodic ceiling and thermodynamic inefficiency of current deep learning, which converges to a statistical average of historic h

BiomedXPro: Prompt Optimization for Explainable Diagnosis with Biomedical Vision Language Models

arXiv:2510.15866v2 Announce Type: replace Abstract: The clinical adoption of biomedical vision-language models is hindered by prompt optimization techniques that produce either uninterpretable latent

safetyarxiv-cs-cv

Brain foundation model-guided source-selective domain adaptation for cross-subject EEG decoding

arXiv:2507.21037v3 Announce Type: replace Abstract: Cross-subject motor-imagery electroencephalography (MI-EEG) decoding remains challenging because substantial inter-subject variability can cause bot

BRANCH: Bypassing Multi-Scanner AI Guardrails

arXiv:2610.10742v1 Announce Type: cross Abstract: AI systems increasingly rely on Large Language Models (LLMs) as core reasoning engines, making them targets for prompt injection and jailbreaks. Guard

BridgeGuard: Explicit Safety Drift for Diffusion-based Autonomous Driving

arXiv:2610.11483v1 Announce Type: new Abstract: Diffusion-based driving planners capture diverse behaviors but can generate unsafe trajectories under distribution shift. We propose BridgeGuard, a safe

CausalDreamer: Learning Predictive World Models with Latent Disentanglement

arXiv:2610.12016v1 Announce Type: new Abstract: World models for control must capture which aspects of the environment respond to the agent's actions and which are relevant to reward. Generative world

Closing the Horizon Gap in Policy Optimization for Adversarial MDPs

arXiv:2610.12362v1 Announce Type: new Abstract: We consider policy optimization for online episodic tabular Markov decision processes (MDPs) with adversarial losses and bandit feedback. Policy optimiz

Comparing Object Detection Models for Electrical Substation Component Mapping

arXiv:2512.22454v2 Announce Type: replace Abstract: Electrical substations are a significant component of an electrical grid. Indeed, the assets at these substations (e.g., transformers) are vulnerabl

safetyarxiv-cs-cv

Connected Self Forcing: Beyond Local Learning in Video Autoregression

arXiv:2610.12156v1 Announce Type: new Abstract: To stream long videos while maintaining visual quality and temporal consistency, Self Forcing mitigates exposure bias through self-rollout training on s

safetyarxiv-cs-cv