Search: “safety”

Robust Optimization for Mitigating Reward Hacking with Correlated Proxies

arXiv:2604.12086v1 Announce Type: new Abstract: Designing robust reinforcement learning (RL) agents in the presence of imperfect reward signals remains a core challenge. In practice, agents are often

15 Apr 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

TEMPLATEFUZZ: Fine-Grained Chat Template Fuzzing for Jailbreaking and Red Teaming LLMs

DGX agent

arXiv:2604.12232v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly deployed across diverse domains, yet their vulnerability to jailbreak attacks, where adversarial inputs

safetyarxiv-cs-ai

15 Apr 2026

Understanding and Improving Continuous Adversarial Training for LLMs via In-context Learning Theory

arXiv:2604.12817v1 Announce Type: new Abstract: Adversarial training (AT) is an effective defense for large language models (LLMs) against jailbreak attacks, but performing AT on LLMs is costly. To im

15 Apr 2026

Anthropogenic Regional Adaptation in Multimodal Vision-Language Model

arXiv:2604.11490v1 Announce Type: new Abstract: While the field of vision-language (VL) has achieved remarkable success in integrating visual and textual information across multiple languages and doma

Model Releases

Consistency of AI-Generated Exercise Prescriptions: A Repeated Generation Study Using a Large Language Model

model-releasesarxiv-cs-ai

arXiv:2604.11287v1 Announce Type: new Abstract: Background: Large language models (LLMs) have been explored as tools for generating personalized exercise prescriptions, yet the consistency of outputs

Efficient KernelSHAP Explanations for Patch-based 3D Medical Image Segmentation

arXiv:2604.11775v1 Announce Type: cross Abstract: Perturbation-based explainability methods such as KernelSHAP provide model-agnostic attributions but are typically impractical for patch-based 3D medi

Endogenous Information in Routing Games: Memory-Constrained Equilibria, Recall Braess Paradoxes, and Memory Design

arXiv:2604.11733v1 Announce Type: cross Abstract: We study routing games in which travelers optimize over routes that are remembered or surfaced, rather than over a fixed exogenous action set. The pap

Enhancing Fine-Grained Spatial Grounding in 3D CT Report Generation via Discriminative Guidance

arXiv:2604.10437v1 Announce Type: new Abstract: Vision--language models (VLMs) for radiology report generation (RRG) can produce long-form chest CT reports from volumetric scans and show strong potent

safetyarxiv-cs-cv

Examining EAP Students' AI Disclosure Intention: A Cognition-Affect-Conation Perspective

arXiv:2604.10991v1 Announce Type: cross Abstract: The growing use of generative artificial intelligence (AI) in academic writing has raised increasing concerns regarding transparency and academic inte

FREE-Switch: Frequency-based Dynamic LoRA Switch for Style Transfer

arXiv:2604.10023v1 Announce Type: cross Abstract: With the growing availability of open-sourced adapters trained on the same diffusion backbone for diverse scenes and objects, combining these pretrain

Learning to Focus: CSI-Free Hierarchical MARL for Reconfigurable Reflectors

arXiv:2604.05165v2 Announce Type: replace Abstract: Reconfigurable Intelligent Surfaces (RIS) has a potential to engineer smart radio environments for next-generation millimeter-wave (mmWave) networks

Like a Hammer, It Can Build, It Can Break: Large Language Model Uses, Perceptions, and Adoption in Cybersecurity Operations on Reddit

arXiv:2604.09998v1 Announce Type: cross Abstract: Large language models (LLMs) have recently emerged as promising tools for augmenting Security Operations Center (SOC) workflows, with vendors increasi

MedLVR: Latent Visual Reasoning for Reliable Medical Visual Question Answering

arXiv:2604.09757v1 Announce Type: cross Abstract: Medical vision--language models (VLMs) have shown strong potential for medical visual question answering (VQA), yet their reasoning remains largely te

MOSAIC: Multi-Domain Orthogonal Session Adaptive Intent Capture for Prescient Recommendations

arXiv:2604.10147v1 Announce Type: cross Abstract: Capturing user intent across heterogeneous behavioral domains stands as a fundamental challenge in session-based recommender systems. Yet, existing mu

Multi-Model Synthetic Training for Mission-Critical Small Language Models

arXiv:2509.13047v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities across many domains, yet their application to specialized fields remain

OOM-RL: Out-of-Money Reinforcement Learning Market-Driven Alignment for LLM-Based Multi-Agent Systems

arXiv:2604.11477v1 Announce Type: new Abstract: The alignment of Multi-Agent Systems (MAS) for autonomous software engineering is constrained by evaluator epistemic uncertainty. Current paradigms, suc

PEMANT: Persona-Enriched Multi-Agent Negotiation for Travel

arXiv:2604.10475v1 Announce Type: new Abstract: Modeling household-level trip generation is fundamental to accurate demand forecasting, traffic flow estimation, and urban system planning. Existing stu

Preference-Agile Multi-Objective Optimization for Real-time Vehicle Dispatching

arXiv:2604.10664v1 Announce Type: new Abstract: Multi-objective optimization (MOO) has been widely studied in literature because of its versatility in human-centered decision making in real-life appli

SemaClaw: A Step Towards General-Purpose Personal AI Agents through Harness Engineering

arXiv:2604.11548v1 Announce Type: new Abstract: The rise of OpenClaw in early 2026 marks the moment when millions of users began deploying personal AI agents into their daily lives, delegating tasks r

Sense Less, Infer More: Agentic Multimodal Transformers for Edge Medical Intelligence

arXiv:2604.10404v1 Announce Type: cross Abstract: Edge-based multimodal medical monitoring requires models that balance diagnostic accuracy with severe energy constraints. Continuous acquisition of EC

Steered LLM Activations are Non-Surjective

arXiv:2604.09839v1 Announce Type: new Abstract: Activation steering is a popular white-box control technique that modifies model activations to elicit an abstract change in output behavior. It has als

Towards Adaptive Open-Set Object Detection via Category-Level Collaboration Knowledge Mining

arXiv:2604.11195v1 Announce Type: cross Abstract: Existing object detectors often struggle to generalize across domains while adapting to emerging novel categories. Adaptive open-set object detection

Trajectory-based actuator identification via differentiable simulation

arXiv:2604.10351v1 Announce Type: new Abstract: Accurate actuation models are critical for bridging the gap between simulation and real robot behavior, yet obtaining high-fidelity actuator dynamics ty

Weird Generalization is Weirdly Brittle

arXiv:2604.10022v1 Announce Type: new Abstract: Weird generalization is a phenomenon in which models fine-tuned on data from a narrow domain (e.g. insecure code) develop surprising traits that manifes

safetyarxiv-cs-cl

A Restore Britain Government would back British drivers. How? Raise the speed limit on motorways to 80mph. Remove all 20mph zones, other tha…

A Restore Britain Government would back British drivers. How? Raise the speed limit on motorways to 80mph. Remove all 20mph zones, other than those outside schools or areas with vulnerable individuals

safetyelon-musk--x

Better Eyes, Better Thoughts: Why Vision Chain-of-Thought Fails in Medicine

arXiv:2603.06665v2 Announce Type: replace-cross Abstract: Large vision-language models (VLMs) often benefit from chain-of-thought (CoT) prompting in general domains, yet its efficacy in medical vision

CausalVAD: De-confounding End-to-End Autonomous Driving via Causal Intervention

arXiv:2603.18561v2 Announce Type: replace Abstract: Planning-oriented end-to-end driving models show great promise, yet they fundamentally learn statistical correlations instead of true causal relatio

safetyarxiv-cs-cv

ECHO: Efficient Chest X-ray Report Generation with One-step Block Diffusion

arXiv:2604.09450v1 Announce Type: cross Abstract: Chest X-ray report generation (CXR-RG) has the potential to substantially alleviate radiologists' workload. However, conventional autoregressive visio

Large Language Models Generate Harmful Content Using a Distinct, Unified Mechanism

arXiv:2604.09544v1 Announce Type: cross Abstract: Large language models (LLMs) undergo alignment training to avoid harmful behaviors, yet the resulting safeguards remain brittle: jailbreaks routinely

On the Spectral Geometry of Cross-Modal Representations: A Functional Map Diagnostic for Multimodal Alignment

arXiv:2604.08579v1 Announce Type: cross Abstract: We study cross-modal alignment between independently pretrained vision (DINOv2) and language (all-MiniLM-L6-v2) encoders using the functional map fram

Semantic Intent Fragmentation: A Single-Shot Compositional Attack on Multi-Agent AI Pipelines

arXiv:2604.08608v1 Announce Type: cross Abstract: We introduce Semantic Intent Fragmentation (SIF), an attack class against LLM orchestration systems where a single, legitimately phrased request cause

Is Elon Musk right about Sam Altman? just read the new Ronan Farrow and Andrew Marantz piece on Sam Altman in The New Yorker, and it is damn…

Is Elon Musk right about Sam Altman? just read the new Ronan Farrow and Andrew Marantz piece on Sam Altman in The New Yorker, and it is damning. And the answer is yes. It’s a deep dive based on never-

safetyelon-musk--x

12 Apr 2026

Are GUI Agents Focused Enough? Automated Distraction via Semantic-level UI Element Injection

arXiv:2604.07831v1 Announce Type: cross Abstract: Existing red-teaming studies on GUI agents have important limitations. Adversarial perturbations typically require white-box access, which is unavaila

safetyarxiv-cs-cl

Does anyone without a lot of money at stake seriously disagree?

I was unable to retrieve the specific tweet at the provided URL. The tweet ID `2042429820500656544` does not appear in any indexed search results, and X (formerly Twitter) requires JavaScript/login...

safetygary-marcus--x

Download the Tesla Robotaxi app

Download the Tesla Robotaxi app NEWS: Tesla has updated their Robotaxi app. What's new: • New tips section, which displays how many seats are available in the car, tells you that the vehicle lights wi

safetyelon-musk--x

Learning Without Losing Identity: Capability Evolution for Embodied Agents

arXiv:2604.07799v1 Announce Type: new Abstract: Embodied agents are expected to operate persistently in dynamic physical environments, continuously acquiring new capabilities over time. Existing appro

Machine Unlearning in the Era of Quantum Machine Learning: An Empirical Study

arXiv:2512.19253v4 Announce Type: replace-cross Abstract: We present the first empirical study of machine unlearning (MU) in hybrid quantum-classical neural networks. While MU has been extensively exp

SymptomWise: A Deterministic Reasoning Layer for Reliable and Efficient AI Systems

arXiv:2604.06375v1 Announce Type: new Abstract: AI-driven symptom analysis systems face persistent challenges in reliability, interpretability, and hallucination. End-to-end generative approaches ofte

The reaction people are having to AIs that can find bugs in code is fascinating. Finally, we have the capacity to fix the crisis in computer…

The reaction people are having to AIs that can find bugs in code is fascinating. Finally, we have the capacity to fix the crisis in computer security we’ve had for decades, and everyone is treating it

safetyclem-delangue--x

If you ban self-driving cars to protect the taxi union, you have blood on your hands

If you ban self-driving cars to protect the taxi union, you have blood on your hands If you want to know why @Waymo is no longer testing in NYC, this statement says it all: “Our top priority for AV te

safetysonya-huang--x

7 Apr 2026

CLIS-MPPI: Closed-Loop Model Predictive Path Integral Control With Initial-State Sampling Under Execution Delay

arXiv:2610.21239v1 Announce Type: new Abstract: Sampling-based motion planning such as Model Predictive Path Integral (MPPI) control has become attractive for robotics due to its ability to handle non

Compliant GMT: Making General Motion Tracking Compliant via Residual Reference Adaptation

arXiv:2610.22119v1 Announce Type: new Abstract: We propose Compliant GMT (C-GMT), a modular framework for full-body compliant responses in general motion tracking. As humanoid robots move into real wo

DemoStream: Batched Task-and-Motion-Plan Optimization for Automated Learning from Demonstration

arXiv:2610.22020v1 Announce Type: cross Abstract: Learning from demonstration is an effective strategy for endowing robots with manipulation capabilities but requires training on datasets comprised of

Evolving Together, Failing Apart: The Skill-Policy Co-evolution Trap in Self-Improving Agents

arXiv:2610.20845v1 Announce Type: new Abstract: Skill-policy co-evolution, in which an agent's policy model and its skill bank improve each other during training, is a promising paradigm for self-impr

Learning Changes the World and Then What Was Learned has Changed: Training and Deploying Neural Networks in the Control Loop

arXiv:2610.22084v1 Announce Type: cross Abstract: Neural networks offer critical adaptability for controlling complex systems where nominal modeling proves intractable. However, when offline-trained n

Learning When to Withdraw: Curriculum Coverage and Rule-Specific Failures in Fine-Tuned Language Agents

arXiv:2610.21045v1 Announce Type: new Abstract: Tool-using language agents must revise prepared actions when evidence changes or authority is revoked. Selective withdrawal means suspending affected ac

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

arXiv:2603.22519v4 Announce Type: replace-cross Abstract: Textual Large Language Models (LLMs) provide a simple and familiar interface: a string of text is used for both input and output. However, the

Manifold Generative Annealing for Contact Constraint-aware Impedance Control with Reinforcement Learning Prior

arXiv:2610.21140v1 Announce Type: new Abstract: Contact-rich robotic tasks require coordinated motion and impedance under uncertain interaction dynamics and physical constraints. While recent methods