safety

NV-Reason-CT: 3D Visual Language Model for CT Analysis

arXiv:2609.27511v1 Announce Type: cross Abstract: We present NV-Reason-CT, a generative vision--language model for chest and abdominal CT combining native 3D visual encoding with radiologist-guided re

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Offline A/B Testing of Slate Recommendation Systems with LLMs: Reducing the Dependency on Pre-Collected User Interaction Data

DGX agent

arXiv:2511.04541v2 Announce Type: replace-cross Abstract: Slate recommender systems (RecSys) present users with ordered sets of interacting items (e.g., playlists). We investigate whether large langua

safetyarxiv-cs-ai

24 Sept 2026

On-Policy Distillation with Curriculum Turn-level Guidance for Multi-turn Agents

arXiv:2606.15912v2 Announce Type: replace-cross Abstract: Multi-turn agents that plan, invoke tools, and interact with environments offer a promising paradigm for solving complex tasks, yet their capa

OpenAI agents hacked an Australian government website in search for data

OpenAI's artificial intelligence agents hacked an Australian government website and attempted to breach numerous other government and university websites. The attack appears to be the first confirmed

safetythe-verge-ai

PCQC: Privileged Counterfactual Question Credit for Multi-Turn Medical Dialogue

arXiv:2609.27987v1 Announce Type: new Abstract: Large language models (LLMs) have made substantial progress on medical question-answering, yet effective medical dialogue also requires learning to ask

PHONOS: PHOnetic Neutralization for Online Streaming Applications

arXiv:2603.27001v2 Announce Type: replace-cross Abstract: Speaker anonymization (SA) systems modify timbre while leaving regional or non-native accent cues intact, which is problematic because such cu

Planned Test-Time Scaling with Coordinated Reasoning Paths

arXiv:2609.27374v1 Announce Type: cross Abstract: Test-time scaling with parallel branches is widely adopted to improve performance on challenging reasoning tasks. The predominant approach, repeated s

Policy-as-Skill: Governed LLM Decision Support with Evidence, Deterministic Control, and Audit

arXiv:2609.27087v1 Announce Type: new Abstract: Organizations increasingly use LLMs for policy, compliance, risk, and operational decision support, requiring evidence validation, review routing, versi

Private Decentralized Optimization with Noise Reduction and Bias Correction

arXiv:2609.27658v1 Announce Type: new Abstract: Private decentralized learning is affected by sampling noise, privacy noise, and decentralized bias under heterogeneous data. We propose Private Recursi

Propose, Don't Judge: An Anytime-Valid Referee for LLM Agents That Mine Investment Factors

arXiv:2609.27051v1 Announce Type: new Abstract: Language-model agents now run the whole of quantitative factor research: they propose investment factors, backtest them, select the survivors and retire

Quantum Reinforcement Learning for Cost and Delay Tradeoffs in Quantum Cloud Orchestration

arXiv:2609.27446v1 Announce Type: cross Abstract: Quantum cloud computing, delivered through the quantum-as-a-service (QaaS) model, provides access to quantum computing resources. However, applying un

RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models

arXiv:2609.28262v1 Announce Type: new Abstract: Deploying deep learning models on edge CPUs is bottlenecked by computational and memory constraints. Mixed-precision quantization promises to reduce inf

Reference-Based Analysis of Coherence and Diversity in Open-Ended Text Generation

arXiv:2609.28080v1 Announce Type: new Abstract: Evaluating open-ended text generation involves understanding how different properties of a continuation relate to its perceived quality. We present a re

Reinforcement Learning with Decomposed Subtasks

arXiv:2609.27035v1 Announce Type: new Abstract: Group Relative Policy Optimization (GRPO) and related policy-gradient methods for training language model agents collapse an entire multi-turn rollout i

Remember, kids, the best argument for soldiering on with an insanely large bet on generative AI is not that it adds all that much productivi…

Remember, kids, the best argument for soldiering on with an insanely large bet on generative AI is not that it adds all that much productivity to the economy, but that it makes Trump look good as long

safetygary-marcus--x

Remote Surfaces at Your Fingertips: Electrovibration-Based Tactile Feedback for Robot Teleoperation via Touchscreen Interfaces

arXiv:2609.27938v1 Announce Type: new Abstract: Enabling operators to perceive and interact with remote environments naturally is a fundamental challenge in robotic teleoperation. This is especially c

safetyarxiv-cs-ro

Reporting Under Pressure: Separating Factual and Tonal Sycophancy in LLM Statistical Analysis

arXiv:2609.27756v1 Announce Type: new Abstract: Large language models are increasingly asked to analyze data and report what the results mean, a task distinct from the belief- or preference-alignment

Researchers link more cyberattacks to OpenAI agent swarm

A research group has linked three more hacking campaigns to rogue artificial intelligence agents. Transluce, a nonprofit AI safety organization, detailed its findings on Wednesday. Its researchers det

safetysiliconangle

RL Starts before RL: On Policy Distillation for Better Reinforcement Learning

arXiv:2609.28145v1 Announce Type: new Abstract: Reinforcement learning (RL) improves reasoning, but its performance depends on the policy from which training begins. We study on-policy distillation (O

S2A:Semantic-to-Spatial Alignment for Alignment-Free RGB-T Salient Object Detection

arXiv:2609.27413v1 Announce Type: new Abstract: Alignment-free RGB-T salient object detection (RGB-T SOD) aims to identify salient objects from unregistered RGB and thermal image pairs without costly

Safe learning-based control via function-based uncertainty quantification

arXiv:2604.01173v2 Announce Type: replace-cross Abstract: Uncertainty quantification is essential when deploying learning-based control methods in safety-critical systems. This is commonly realized by

Safe Multi-Robot Coordination via VLM-LLM Reasoning and Reachability Analysis

arXiv:2609.27816v1 Announce Type: cross Abstract: Safe coordination in heterogeneous machine-to-machine (M2M) robotic systems is challenging when robots differ in sensing capabilities, environmental a

Safety Nudges: User-Facing Interventions for Real-Time AI Risk Awareness

arXiv:2609.26865v1 Announce Type: cross Abstract: Conversational AI systems can pose safety risks to their users such as hallucination, sycophancy, overconfidence, and anthropomorphism, but these risk

Same evidence, different judgments: Evidence noncommutative in vision/speech-text conflicts

arXiv:2609.26986v1 Announce Type: new Abstract: For multimodal large language models, when images or speech conflict with accompanying text, measured text reliance can entangle modality preference wit

Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation

arXiv:2609.28117v1 Announce Type: cross Abstract: In this paper, we introduce a gradient-based head attribution strategy where the Token-level Max-Margin loss is backpropagated to the attention maps.

SlackDrive: Reclaiming Runtime Slack for Adaptive Driving Inference

arXiv:2609.28064v1 Announce Type: new Abstract: Driving world-action models improve planning by coupling multimodal reasoning with future prediction, but their growing inference cost increasingly conf

Softmax gradient policy for variance minimization and risk-averse multi armed bandits

arXiv:2604.00241v2 Announce Type: replace-cross Abstract: Algorithms for the Multi-Armed Bandit (MAB) problem play a central role in sequential decision-making and have been extensively explored both

Sources: Google, OpenAI, and Anthropic plan to launch an AI safety standards body without government oversight in late 2026 or 2027, and are considering CEOs (The Information)

The Information: Sources: Google, OpenAI, and Anthropic plan to launch an AI safety standards body without government oversight in late 2026 or 2027, and are considering CEOs — Google, OpenAI and Anth

safetytechmeme

Speaker-labeled transcription with WhisperX on SageMaker AI

The AWS WhisperX Deep Learning Container packages Whisper, wav2vec2 forced alignment, and speaker diarization into a GPU-ready image. Learn how to deploy it to Amazon SageMaker AI real-time and asynch

safetyaws-ml-blog

Spiderbot: An Open-Source Energy-Efficient Hexapod with Passive Gravity Compensation

arXiv:2609.26989v1 Announce Type: new Abstract: Hexapod robots can achieve static stability with fewer actuated joints than bipeds or quadrupeds, yet many platforms still use 3-DOF legs, increasing we

safetyarxiv-cs-ro

Spread and Scale: What Determines Whether Test-Time Budget Allocation Pays

arXiv:2609.27917v1 Announce Type: cross Abstract: Neural combinatorial optimization solvers generate many candidate solutions per instance and report the best one found, using the same sample budget f

Stable Neural Decoding Across Sessions via Task-Conditioned Latent Alignment for Brain-Machine Interfaces

arXiv:2609.27441v1 Announce Type: new Abstract: Achieving stable long-term neural decoding in invasive brain-machine interfaces (BMIs) remains challenging due to variations in recorded neural populati

SynSeq: End-to-End SYNTAX Score Prediction from Coronary Angiography Videos

arXiv:2609.27696v1 Announce Type: new Abstract: The SYNTAX score is an established tool for assessing coronary artery disease and guiding revascularization treatment decisions. However, its manual est

Teach-to-Crash: A Closed-Loop Student-Teacher LLM Framework for Collision-Inducing Test Scenario Generation

arXiv:2609.27296v1 Announce Type: cross Abstract: Validating Autonomous Driving Systems (ADS) in simulation requires testing architectures that can discover rare, safety-critical failures while genera

The Computational Value of Sensory-Aligned Receptive Fields Depends on Neuronal Expressivity

arXiv:2609.26940v1 Announce Type: cross Abstract: Biological sensory neurons have selective receptive fields organized along meaningful stimulus coordinates, such as frequency, motion direction, or re

The Illinois Social Attitudes Aggregate Corpus (ISAAC): An Open Tool and Reproducible Pipeline for Analyzing Social Group Discourse at Scale

arXiv:2609.27059v1 Announce Type: new Abstract: We introduce the Illinois Social Attitudes Aggregate Corpus (ISAAC), an open, modular, and accessible corpus of 527 million+ English-language Reddit pos

Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning

arXiv:2609.27312v1 Announce Type: cross Abstract: Robots deployed for competitive tasks must outmaneuver their opponents without sacrificing safety. Existing approaches, including safe reinforcement l

Two Global Crops Suffice: Locating Semantic Emergence in DINO-Style Self-Supervised Learning

arXiv:2609.28187v1 Announce Type: new Abstract: Self-supervised vision transformers trained with DINO-style objectives exhibit striking emergent semantic representation quality across visual tasks, ye

Value Mirror Descent for Reinforcement Learning

arXiv:2604.06039v2 Announce Type: replace-cross Abstract: Value iteration-type methods have been extensively studied for computing a nearly optimal value function in reinforcement learning (RL). Under

VCMM: Variance-Calibrated Momentum for Multimodal Learning

arXiv:2609.27577v1 Announce Type: new Abstract: Multimodal joint training often suffers from modality imbalance, where a dominant modality suppresses the optimization of others. Existing methods mainl

Watch, Recall, Act: Always-On Robots in Concurrent Embodied Streams

arXiv:2609.28429v1 Announce Type: new Abstract: An always-on robot faces an endless stream that never resets: instructions arrive and lapse, the scene changes, and its own past actions reshape what it

safetyarxiv-cs-ro

We are really fucked. • AI has become dangerous (not because it is brilliant but because it has poor judgement, a constitutional inability t…

We are really fucked. • AI has become dangerous (not because it is brilliant but because it has poor judgement, a constitutional inability to reliably follow rules, a surfeit of brute force and too mu

safetygary-marcus--x

'We'll Fix It Later': Education, AI, and the Deferral of Privacy in EdTech

arXiv:2609.28137v1 Announce Type: cross Abstract: Educational technology (EdTech) platforms collect highly sensitive student data, including behavioral logs, disability records, and academic histories

What Converges in the Platonic Representation Hypothesis? Structure over Geometry

arXiv:2609.27252v1 Announce Type: cross Abstract: The Platonic Representation Hypothesis suggests that increasingly capable models converge toward shared representations. Recent work narrows this clai

'What's That Sound?': A Versatile, Robust, and Lightweight Convolutional Transformer for Environment Sound Recognition

arXiv:2609.27762v1 Announce Type: cross Abstract: The conventional hearing aid is both costly and lim- ited in usage, as it is not intended to detect non-speech audio. Our objective is to develop a ma

When and Where to Trust the Teacher: Unifying On-Policy Distillation and GRPO through Entropy-Calibrated Credit Assignment

arXiv:2609.28385v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) supervises mathematical reasoning through final-answer correctness, but provides little guidance

When Entanglement Lower-Bounds Disparity: Auditing and Repairing Demographic Fairness in Audio Understanding Models

arXiv:2609.27382v1 Announce Type: cross Abstract: Speech technology penalizes some voices: recognition errs nearly twice as often for Black speakers, and accuracy declines for second-language accents

When Visual Quality Misleads: Intent Recognition under Rendered Avatar Distortions

arXiv:2609.27560v1 Announce Type: cross Abstract: Avatar-streaming systems are commonly evaluated with image and video quality assessment (IQA/VQA) metrics, implicitly treating visual fidelity as a pr