safety

Semantically Labelled Automata for Multi-Task Reinforcement Learning with LTL Instructions

arXiv:2602.06746v2 Announce Type: replace Abstract: We study multi-task reinforcement learning (RL), a setting in which an agent learns a single, universal policy capable of generalising to arbitrary,

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Sensor-Driven Mission Synthesis for UAV/UGV Swarms: A TB-CSPN Coordination Architecture with Hardware-Enforced Safety

DGX agent

arXiv:2608.14306v1 Announce Type: new Abstract: This paper presents a coordination architecture for heterogeneous UAV/UGV swarms that synthesises mission actions from uncertain, multi-modal sensor evi

safetyarxiv-cs-ai

17 Aug 2026

Some say dogs look like their owners. Generative AI CEOs have come to sound like their product: frequently wrong, never in doubt, often full…

Some say dogs look like their owners. Generative AI CEOs have come to sound like their product: frequently wrong, never in doubt, often full of BS. Dario is full of sh@t. Birds of a feather. JUST SAY

Some thoughts on Dario’s post: 1. Dario does not actually address Gavin Baker’s account of what he said – something he could easily deny if …

Some thoughts on Dario’s post: 1. Dario does not actually address Gavin Baker’s account of what he said – something he could easily deny if it were inaccurate. 2. Dario claims his critics live in a “b

safetyyann-lecun--x

Spatiotemporal Tube-Based Safety-Certificate for Autonomous Navigation of Articulated Vehicles

arXiv:2608.14531v1 Announce Type: new Abstract: Articulated vehicles are the workhorses of freight transportation, and their autonomous navigation is challenging. Their physical characteristics and mo

SSP: An Event-Matched Syn2Sim2Phy Cross-Domain Evaluation Framework for Autonomous Driving VLA Models

arXiv:2608.14024v1 Announce Type: new Abstract: Vision-language-action (VLA) models for autonomous driving jointly produce scene interpretation, language-based reasoning, and driving trajectories. Exi

Stochastic Control Policies for Robust Molecular Transition Path Sampling

arXiv:2608.13800v1 Announce Type: new Abstract: Transition path sampling (TPS) aims to efficiently generate rare molecular transition trajectories between metastable states and is essential for unders

Structure-Guided Spatiotemporal Attention Graph Neural Network for Traffic Flow Prediction

arXiv:2608.14177v1 Announce Type: cross Abstract: Deep spatiotemporal models integrating graph convolutions and attention mechanisms have demonstrated excellent performance in network-level traffic fl

The Linear Geometry of Interpretable Tokens: Jailbreaking Attacks and Defenses for Unlearned Diffusion Models

arXiv:2504.21307v3 Announce Type: replace Abstract: Diffusion models excel at generating high-quality images but can memorize and reproduce harmful concepts when prompted. Although fine-tuning methods

Tracking the Flight: Exploring a Computational Framework for Analyzing Escape Responses in Plains Zebra (Equus quagga)

arXiv:2505.16882v3 Announce Type: replace Abstract: Ethological research increasingly benefits from the growing affordability and accessibility of drones, which enable the capture of high-resolution f

Training Fair Tabular Foundation Models

arXiv:2608.14211v1 Announce Type: cross Abstract: Tabular Foundation Models (TFMs) have emerged as leading methods for tabular predictive tasks, leveraging in-context learning to predict on new data w

Training-Free Knowledge Transfer Across Model Scales through Activation-Guided Pruning

arXiv:2608.13596v1 Announce Type: cross Abstract: Heterogeneous model fusion seeks to combine models that differ in tasks, initializations, architectures, or scales. We study an underexplored cross-sc

Tripwire: Triggering Aligned Refusal via Statistically Certified Safety Neurons

arXiv:2608.14392v1 Announce Type: new Abstract: Neuron- and path-level interventions offer the finest-grained route to defending large language models (LLMs) against jailbreak attacks, yet existing me

Understanding and Mitigating Over-refusal for Large Language Models via Representation Intervention

arXiv:2511.19009v2 Announce Type: replace-cross Abstract: Large language models (LLMs) demonstrate powerful capabilities across various natural language processing tasks,yet their inherent safety vuln

safetyarxiv-cs-cl

VoiceDesigner: Text-to-Voice Generation and Editing via Unified Diffusion Modeling and Data Augmentation

arXiv:2608.13613v1 Announce Type: cross Abstract: Recent breakthroughs in generative models have made text-to-voice generation (TTV) possible, enabling the synthesis of speech directly from textual vo

Why we need an AI-resilient society

arXiv:1912.08786v3 Announce Type: replace-cross Abstract: Three generations of software have transformed the role of artificial intelligence in society. In the first, programmers wrote explicit logic.

XSA-MAD: Cross-modal Semantic Alignment for Morphing Attack Detection

arXiv:2608.13861v1 Announce Type: new Abstract: Morphing attacks pose a serious threat to face recognition systems. However, existing image-based morphing attack detection (MAD) methods often generali

JUST IN: OpenAI shuts down team responsible for assessing whether its AI models could pose 'catastrophic risks,' FT reports.

OpenAI has shut down the internal team that evaluated whether its AI models could pose “catastrophic risks,” a decision reported by the Financial Times on August 16, 2026. The move was announced via a

Rogue AI aren’t science fiction anymore

This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on AI safety, follow Robert Hart. The Stepback arrives in our subscribers' inboxes at 8AM ET.

safetythe-verge-ai

@sethharpesq OpenAI spent $1.60 for every dollar it earned in 2025. We need more journalism that cuts through all of the accounting gimmicks…

@sethharpesq OpenAI spent $1.60 for every dollar it earned in 2025. We need more journalism that cuts through all of the accounting gimmicks to get to the heart of what is really happening in the US A

Somebody needs to do an internship at the FDA.

Dr. Catharine Young noted on Twitter that someone should pursue an internship at the U.S. Food & Drug Administration (FDA). On August 16, Polymarket reported that Anthropic CEO Dario Amodei predicted

1/2 Thanks Gavin for an especially thoughtful exchange. I don't usually spend much time on social media but I wanted to engage here because …

1/2 Thanks Gavin for an especially thoughtful exchange. I don't usually spend much time on social media but I wanted to engage here because it really brings out the heart of an important conversation.

safetydario-amodei--x

2/2 Second, on the messaging around AI. I do not agree that my messaging has been disproportionately negative. In fact it has been about equ…

2/2 Second, on the messaging around AI. I do not agree that my messaging has been disproportionately negative. In fact it has been about equally balanced between risks and benefits: I’ve written one m

safetydario-amodei--x

Sources: OpenAI's repeated exec reshuffles and departures have frustrated some staff as it prepares for an IPO; it disbanded its 'preparedness' team in July (Financial Times)

Financial Times: Sources: OpenAI's repeated exec reshuffles and departures have frustrated some staff as it prepares for an IPO; it disbanded its “preparedness” team in July — Executive exits and safe

safetytechmeme

The FT reports this weekend that OpenAI quietly disbanded its preparedness team at the end of last month, with responsibilities for individu…

The FT reports this weekend that OpenAI quietly disbanded its preparedness team at the end of last month, with responsibilities for individual areas of the preparedness framework distributed across ot

Wrong! If an AI leverages symbolic operations (conditionals, operations over variables, code interpreters, etc) and neural networks it is ne…

Wrong! If an AI leverages symbolic operations (conditionals, operations over variables, code interpreters, etc) and neural networks it is neurosymbolic. If it doesn’t, it’s not. ( I laid all of this o

$3 trillion in capex into a technology that has zero correlation with value creation what could possibly go wrong?

$3 trillion in capex into a technology that has zero correlation with value creation what could possibly go wrong? wow! new OpenAI study shows “no correlation between how often employees use AI and ho

A Browser-Native Digital Test Range for Benchmarking 4D Ocean-Glider Planning Algorithms

arXiv:2608.13511v1 Announce Type: new Abstract: Repeated in-situ evaluation of ocean-glider planners requires scarce vehicles, operators, deployment and recovery resources, and ocean conditions that c

A Compositional Theory of Curvature in Probabilistic Circuits

arXiv:2608.12869v1 Announce Type: cross Abstract: Probabilistic Circuits (PCs) are generative models that support exact inference and, unlike deep neural networks, admit an exact and tractable measure

Active-Trace Complexity Bounds for Moreau--Yosida Unadjusted Langevin Sampling

arXiv:2608.13467v1 Announce Type: new Abstract: We study the Moreau--Yosida unadjusted Langevin algorithm (MYULA) for the nonsmooth composite target [ pi(dx)propto exp{-f(x)-g(x)},dx, qquad xinmathbb

AI-Driven Multiscenario Interest Rate Forecasting: A Proof of Concept for Banking Asset Management

arXiv:2608.12424v1 Announce Type: cross Abstract: This study focuses on developing an AI-supported prototype for multiperspective interest rate forecasting that combines classical econometric models w

APEX: Learning Adaptive High-Platform Traversal for Humanoid Robots

arXiv:2602.11143v3 Announce Type: replace Abstract: Humanoid locomotion has advanced rapidly with deep reinforcement learning (DRL), enabling robust feet-based traversal over uneven terrain. Yet platf

ARAC: Benchmarking Auto-Research's Alignment and Completeness on End-to-End Researchs

arXiv:2608.12788v1 Announce Type: new Abstract: The rapid advancement of Auto-Research has surfaced a fundamental evaluation challenge: how can we measure the alignment, logical coherence, and evoluti

Attention from Action, for Action: Emergent Visual Bottlenecks for Policy Learning

arXiv:2608.13422v1 Announce Type: new Abstract: Visual bottlenecks that focus policy inputs on regions of interest (ROIs) can improve data-efficient visuomotor learning by separating where to look fro

Automated Design Optimization via Strategic Search with Large Language Models

arXiv:2511.22651v2 Announce Type: replace-cross Abstract: Optimization methods have long advanced many fields, yet they struggle when faced with design problems where the search space and design param

Better Decomposition, Free Aggregation: A Synthesizer-Folding Framework for Multilingual Multi-Hop Question Answering

arXiv:2608.13160v1 Announce Type: cross Abstract: Multilingual retrieval-augmented generation (mRAG) equips large language models with access to globally distributed external knowledge for complex mul

Beyond Outcome Rewards: Step-Level Self-Distilled Policy Optimization for Deep Search Agents

arXiv:2608.12764v1 Announce Type: cross Abstract: Deep search agents operate over trajectories spanning dozens of steps, yet standard reinforcement learning provides only a single outcome reward per t

Bias Mitigation in Face Recognition via Demographic-based Supervised Contrastive Learning

arXiv:2608.12971v1 Announce Type: new Abstract: Face recognition systems have been shown to be biased toward certain demographic groups by exhibiting different error rates across gender, age, or ethni

Branch and Bound for Relational Verification of Neural Networks

arXiv:2608.13118v1 Announce Type: new Abstract: Verification of neural networks against relational specifications, such as global robustness, is crucial for safety-critical applications of cyber-physi

Chance-constrained selection of sequential intervention strategies from counterfactual estimates

arXiv:2608.13209v1 Announce Type: cross Abstract: Many operational decisions are sequences of interventions under a cumulative resource limit, such as a maintenance schedule within a crew-hour budget.

ContactGuard: Pre-Contact Execution Monitoring with Action-Conditioned Latent World Models

arXiv:2608.13438v1 Announce Type: cross Abstract: Contact-rich manipulation failures are often detected only after the robot has committed to contact. This is especially limiting in wrist-camera setup

Context-Matched Distillation: Teacher Causality for Autoregressive Video Distillation

arXiv:2608.13391v1 Announce Type: new Abstract: Interactive autoregressive video generation demands both low-latency rollouts and precise online control. Few-step distillation accelerates generation b

CROP: Task Relevance via Counterfactuals for Selective On-Policy Distillation

arXiv:2608.13387v1 Announce Type: new Abstract: On-policy distillation (OPD) supervises a student language model on trajectories sampled from its current policy, but assigns equal credit to response t

safetyarxiv-cs-cl

CW-BASS v2: Saturation-Aware Pseudo-Label Selection for Semi-Supervised Segmentation under Foundation-Model Teachers

arXiv:2608.12773v1 Announce Type: new Abstract: Semi-supervised semantic segmentation has long turned on one question, which pseudo-labels to trust, and a generation of selection rules, dynamic thresh

DAPD: Dual-Anchored Policy Distillation

arXiv:2608.01735v2 Announce Type: replace Abstract: On-policy (self) distillation (OPSD) is increasingly adopted for language-model post-training. It strengthens the teacher with privileged informatio

Decoding Task Progress from VLA Representations

arXiv:2608.13474v1 Announce Type: new Abstract: Vision-language-action models (VLAs) are moving rapidly towards deployment as general-purpose manipulation policies, but we currently lack basic tools f

Decoupled Contrastive Decoding via Expert-Aligned Drafting

arXiv:2608.12913v1 Announce Type: new Abstract: Contrastive Decoding (CD) improves generation quality, but its amateur-model pass makes decoding expensive. Accelerating CD with speculative decoding ra

safetyarxiv-cs-cl

DiffGRM: Diffusion-based Generative Recommendation Model

arXiv:2510.21805v2 Announce Type: replace-cross Abstract: Generative recommendation (GR) is an emerging paradigm that represents each item via a tokenizer as an n-digit semantic ID (SID) and predicts