safety

CRAFT: Causal Responsibility and Failure Tracing in Medical Vision Language Models

arXiv:2609.38810v1 Announce Type: cross Abstract: As vision language models are increasingly deployed in clinical diagnosis, understanding how they internally resolve competing visual and textual sign

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Credible scientists don’t estimate things that could conceivably happen, however unlikely, at exactly zero probability.

DGX agent

Credible scientists don’t estimate things that could conceivably happen, however unlikely, at exactly zero probability. AI ‘godfather’ Yann LeCun has ‘zero concerns’ about human extinction, says Anthr

safetygary-marcus--x

1 Oct 2026

Cue the Flow: Steering Flow-Matching Policies for Open-World Delivery Manipulation

arXiv:2609.38989v1 Announce Type: new Abstract: Open-world goods delivery requires mobile manipulators to follow free-form user instructions and manipulate potentially novel objects. Existing dual-sys

Cycle-Aware Autoencoder with Cross-SignalConsistency for Railway Door Anomaly Detection

arXiv:2609.39035v1 Announce Type: cross Abstract: Passenger access doors are safety-critical subsystems in railway vehicles, yet detecting abnormal door behavior in real operation is challenging becau

DAMPER: Return-Prioritized Gradient Control for Smooth Policies

arXiv:2609.38903v1 Announce Type: cross Abstract: Actor-critic methods achieve strong performance in continuous control, but their policies can produce highly oscillatory actions. A common remedy is t

DashVMC: Real-Time Discrete World Model Control in Geometry Dash

arXiv:2609.40003v1 Announce Type: new Abstract: World-model agents are usually evaluated in simulators that can wait for the policy; live games impose the opposite constraint, requiring capture, predi

Debias It Yourself: Teaching LLMs Cognitive Bias Mitigation Interventions

arXiv:2609.40124v1 Announce Type: new Abstract: Bias has long been studied in social psychology and cognitive science, where decades of research have produced a body of validated interventions that re

Delta-K: Boosting Multi-Instance Generation via Cross-Attention Augmentation

arXiv:2603.10210v2 Announce Type: replace-cross Abstract: While Diffusion Models excel in text-to-image synthesis, they frequently suffer from catastrophic concept omission when generating complex mul

Demographic Pluralism: Inference-Time Modeling of Pluralistic Human Preference Distributions

arXiv:2609.38555v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used in culturally sensitive settings, where alignment requires representing diverse preferences within po

DensePed-Lite: Quality-Aware Adaptive Detection for Dense Pedestrians under Occlusion

arXiv:2609.39467v1 Announce Type: new Abstract: Pedestrian detection plays a crucial role in computer vision with applications in autonomous driving, surveillance, and public safety. However, real-wor

Derandomizing Dense Binary Hypervector Codebooks for Quantized Scalars

arXiv:2609.38471v1 Announce Type: cross Abstract: Hyperdimensional computing and vector symbolic architectures often represent quantized scalar levels by dense binary codebooks whose level-to-level si

Diagnosing On-Policy Self-Distillation for Reasoning Language Models

arXiv:2609.39118v1 Announce Type: new Abstract: On-policy self-distillation (OPSD) has attracted growing interest as a promising approach to improve the reasoning ability of language models. Without e

Disagreement-Regularized Imitation Learning for Image-Based Continuous Control with Gaussian and Beta Policies

arXiv:2609.38407v1 Announce Type: new Abstract: Purpose: Behavior cloning can accumulate errors when a learned controller visits states outside the demonstrated distribution. This study evaluates whet

DODGER: Safety-Guided Reinforcement Learning for Robot Navigation Among Dynamic Obstacles

arXiv:2609.38873v1 Announce Type: new Abstract: Robots operating in human-centered environments must safely navigate among multiple dynamic obstacles to avoid collisions with people and surrounding in

Does Text Steer Neural PDE Surrogates? A Controlled Diagnostic with OperatorCLIP

arXiv:2609.38517v1 Announce Type: new Abstract: Lower error from a text-conditioned neural surrogate does not, by itself, show that the model uses the meaning of the text. We examine this attribution

Drape-Compatible Tool-Tip Localization for Hand-Held Laparoscopic Instruments via UWB Carrier-Phase Ranging and Trocar-Constrained Geometry

arXiv:2609.39129v1 Announce Type: new Abstract: A surgical robot policy needs to know where each instrument's working end sits relative to the camera and to the other instrument, yet a laparoscopic op

Drone Soccer: Learning to Manipulate with Multicopter Downwash

arXiv:2609.38588v1 Announce Type: new Abstract: Although multicopter drones are traditionally designed for 'perception-only' tasks, like mapping and exploration, recent work has sought to develop Unma

DualCast: A Dual-Path Language Model for Bimodal Financial Time-Series Forecasting

arXiv:2609.38197v1 Announce Type: cross Abstract: Financial time-series forecasting must capture price dynamics across heterogeneous assets while incorporating news available at prediction time. We in

DynaHarness: A Dynamic Physical Harness for Self-Evolving Robot Agents

arXiv:2609.40306v1 Announce Type: cross Abstract: Pretrained robot policies provide useful action priors, but long-horizon manipulation still requires coordination between semantic reasoning and physi

Efficient Active Auditing of Multi-Group Fairness with Bias Probes

arXiv:2609.40034v1 Announce Type: cross Abstract: Over the past decade, Machine Learning (ML) has been trained under dual objectives: minimizing prediction error via Empirical Risk Minimization (ERM)

Ego4WAM: What Matters When Scaling Egocentric Human Data for Robot Learning?

arXiv:2609.40341v1 Announce Type: cross Abstract: Egocentric human data provides a scalable source of experience for robot learning, but varies substantially in human-robot alignment, behavioral cover

Eigenism: Ethics for a Human-AI Future

arXiv:2606.12420v2 Announce Type: replace-cross Abstract: Our concepts of survival and self-interest were built for single, continuous biological lives. These ideas break down when applied to artifici

EmbodiRSI: Recursive Self-Improvement for Data-Efficient Robot Adaptation

arXiv:2609.38905v1 Announce Type: new Abstract: Adapting robot manipulation policies to new tasks and environments remains highly data-intensive, while the data needed for further improvement depends

Enhancing Autoregressive Video Generation via Representation Adversarial Distillation

arXiv:2609.40037v1 Announce Type: new Abstract: Few-step autoregressive video generation enables efficient streaming synthesis, but errors introduced in early temporal blocks are reused as context and

ETHER: Aligning Emergent Communication for Hindsight Experience Replay

arXiv:2307.15494v3 Announce Type: replace-cross Abstract: Hindsight Experience Replay (HER) enhances sample efficiency in goal-conditioned reinforcement learning (RL) by relabelling failed trajectorie

Evaluating Language Model Safety Across Long Adversarial Conversations

arXiv:2609.38357v1 Announce Type: new Abstract: Conversational safety evaluations often test language models with a single harmful prompt, even though real-world systems interact with users through lo

EvoGuard: An Extensible Agentic RL-based Framework for Practical and Evolving AI-Generated Image Detection

arXiv:2603.17343v3 Announce Type: replace Abstract: The rapid proliferation of AI-Generated Images (AIGIs) poses severe misinformation risks, making AIGI detection critical yet challenging. Traditiona

EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making

arXiv:2609.38334v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as agents for multi-step decision-making, yet transfer poorly to unseen environments. World-model

EWAM: Emergent Depth-Wise Specialization in a Unified Embodied Model -- From Semantic Understanding through Visual Foresight to Action

arXiv:2609.39973v1 Announce Type: new Abstract: Vision-language-action (VLA) policies emphasize semantic understanding, whereas world-action models (WAMs) learn predictive representations of environme

Explicit Trajectory Diversity for RL-Based Post-Training of LLM Agents

arXiv:2609.38805v1 Announce Type: cross Abstract: LLM agents often admit multiple high-quality solutions to the same task, differing in reasoning structure, tool-use pattern, or interaction trajectory

ExploreNet: Learning Where to Explore in Diffusion GRPO

arXiv:2609.38329v1 Announce Type: new Abstract: Group-relative RL methods such as Flow-GRPO post-train image generators by exploring with isotropic Gaussian noise added at every denoising step. This n

Fairness Beyond a Single Run: Training-Seed Variability in Speech LLM Adaptation

arXiv:2609.38976v1 Announce Type: new Abstract: Demographic fairness gaps in automatic speech recognition are almost always reported from a single training run. We fine-tune the Q-former projector and

Fairness Theatre: Evaluating Post-Hoc Fairness Interventions in Vendor-Controlled Early Warning Systems

arXiv:2609.38552v1 Announce Type: cross Abstract: Public institutions increasingly procure AI systems whose design they cannot inspect or change. In higher education, proprietary Early Warning Systems

FairRARI: A Plug and Play Framework for Fairness-Aware PageRank

arXiv:2602.08589v2 Announce Type: replace Abstract: PageRank (PR) is a fundamental algorithm in graph machine learning tasks. Owing to the increasing importance of algorithmic fairness, we consider th

Fast Regularized Policy Mirror Descent with One-Step TD Updates

arXiv:2609.39837v1 Announce Type: new Abstract: Policy mirror descent (PMD) enjoys fast convergence in regularized Markov decision processes (MDPs), but existing guarantees often rely on exact or incr

Fenchel Tilting: Weighted Correction for Efficient Finetuning of Generative Models

arXiv:2609.40030v1 Announce Type: cross Abstract: Adapting a pretrained generative model to an arbitrary preference expressed as a utility function underlies reward alignment, guided design, and const

FLOW: Feature-Level Optimal Warping for Generalized Remote Physiological Measurement

arXiv:2609.38913v1 Announce Type: new Abstract: Remote photoplethysmography (rPPG) enables non-contact physiological measurement but remains vulnerable to domain shifts from illumination, motion, and

FoCLIP: A Feature-Space Misalignment Framework for CLIP-Based Image Manipulation and Detection

arXiv:2511.06947v2 Announce Type: replace-cross Abstract: The well-aligned attribute of CLIP-based models enables its effective application like CLIPscore as a widely adopted image quality assessment

Fork-dLLM: Avoiding the Flexibility Trap in Diffusion Language Models

arXiv:2609.39859v1 Announce Type: new Abstract: Masked diffusion language models (dLLMs) have shown strong potential for faster inference through parallel token generation when combined with confidenc

FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation

arXiv:2609.38839v1 Announce Type: new Abstract: Long-horizon video generation requires models to effectively leverage an increasingly long generation history. As the generated history grows, retaining

From Image Interpretation to Clinical Reasoning: Upstream Physician-Context-Aware Multimodal Learning with Causal Reinforcement Learning

arXiv:2609.38924v1 Announce Type: new Abstract: Major adverse cardiovascular events (MACE) remain the leading cause of mortality worldwide. Opportunistic screening using routinely acquired clinical da

From Imitation to Reward Discovery: On-Policy Warmup for Agentic RL

arXiv:2609.39436v1 Announce Type: cross Abstract: Reinforcement learning with a verifiable reward (RLVR) offers a scalable approach to training language-model agents, yet sparse outcome rewards can le

From Local Whole-Body VLA Behaviors to Scene-Scale Aerial Manipulation

arXiv:2609.39670v1 Announce Type: new Abstract: Vision-language-action (VLA) models enable task-conditioned interaction, but extending them to scene-scale aerial manipulation remains challenging due t

From Speech to Editable Concepts: Probing Emotion Recognition with Concept Bottleneck Models

arXiv:2609.39453v1 Announce Type: cross Abstract: Speech emotion recognition (SER) is the task of assigning emotion labels to utterances. Early systems relied on acoustic features, whereas recent appr

From Tokens to Policy: Causal and Interpretable Heterogeneous Treatment Effects Identification

arXiv:2606.17010v2 Announce Type: replace Abstract: Heterogeneous Treatment Effect (HTE) identification is essential to explain the impact of an intervention and optimize our policies accordingly. Exi

From Wrecks to Wisdom: Recovering Crash Mechanics from Real-World Multi-View Photos

arXiv:2609.39486v1 Announce Type: new Abstract: Estimating accident mechanics from real-world crashes is important for vehicle-safety analysis, injury modeling, crash-severity prediction, and operatio

Future Video Generation Better Aligns with the Human Visual Cortex than Observed Video

arXiv:2609.38819v1 Announce Type: cross Abstract: Studying the alignment between the internal representations of vision models and the responses of the visual cortex to the same observed visual stimul

Game-Guided Skill Discovery through Self-Play for Playable Agent Control

arXiv:2609.40137v1 Announce Type: cross Abstract: We present Game-Guided Skill Discovery (GGSD), a framework that uses self-play in games to discover motor skills that are directly playable by humans.