safety

Entropy Regularization: A Free Correction to Cross-Entropy for Verified Demonstrations

arXiv:2609.30572v1 Announce Type: new Abstract: Large language models are often post-trained on expert demonstrations using cross-entropy (CE), even when the downstream objective is not to imitate the

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Evaluation Is All You Need for Multi-Modal Autonomous Driving

DGX agent

arXiv:2609.30818v1 Announce Type: cross Abstract: Multi-modal planning is promising for autonomous driving by representing multiple plausible behaviors in ambiguous and long-tail scenarios. Existing m

safetyarxiv-cs-ai

28 Sept 2026

Evolutionary Safety of Recursive Self-Improving AI: Taxonomy, Risk Discovery, and Evaluation

arXiv:2609.31186v1 Announce Type: new Abstract: Artificial intelligence is advancing rapidly, with increasingly capable systems taking larger roles in reasoning, decision-making, scientific discovery,

Exploring the Benefits of Vision Foundation Models for Unsupervised Domain Adaptation

arXiv:2406.09896v3 Announce Type: replace Abstract: Achieving robust generalization across diverse data domains remains a significant challenge in computer vision. This challenge is important in safet

safetyarxiv-cs-cv

FARE: Forensic Acceptance Region Estimation for Catching Bait-and-Switch Image Generators

arXiv:2609.30982v1 Announce Type: cross Abstract: Modern AI image generators are increasingly deployed as opaque APIs, where customers can query the deployed service, but cannot inspect model weights

Feeding BabyLMs Macaroni: Code-Switching Curricula Cause Cross-Lingual Convergence

arXiv:2609.30535v1 Announce Type: new Abstract: Children in multilingual communities often code-switch, using multiple languages in a single utterance. Can we induce cross-lingual alignment in languag

safetyarxiv-cs-cl

Financial Fragility in Societies of LLM Agents: Coordination Failures and Stabilizing Mechanisms

arXiv:2609.30940v1 Announce Type: new Abstract: Individually protective decisions can produce avoidable collective failures. As large language model (LLM) agents take on greater roles in financial dec

FLAME: Flow Enhanced Legendre Memory Models for General Time Series Forecasting

arXiv:2512.14253v4 Announce Type: replace Abstract: In this work, we introduce FLAME, a family of extremely lightweight and capable Time Series Foundation Models, which support versatile forecasting t

FLIP: Final Layer Inference-Time Probing for Vision-Language Models

arXiv:2609.30993v1 Announce Type: cross Abstract: We present FLIP, a final-layer inference-time probe for testing whether a logit-facing intervention site in an open-weight vision-language model (VLM)

Florida gets it: we need an injunction NOW, otherwise we are playing Russian roulette, hoping that some new software package can save us bef…

Florida gets it: we need an injunction NOW, otherwise we are playing Russian roulette, hoping that some new software package can save us before some massive and possibly deadly AI-induced infrastructu

safetygary-marcus--x

Florida seeks a ban on ChatGPT acting like a person

Florida Attorney General James Uthmeier is calling for a judge to block OpenAI from 'giving ChatGPT false human attributes,' a few months after Florida sued the AI company over safety concerns. Accord

safetythe-verge-ai

FRAM: Trajectory-Guided Visual Feature Selection for Compact Language-Conditioned Robot Manipulation

arXiv:2609.30965v1 Announce Type: new Abstract: Vision-Language-Action models achieve strong performance in robot manipulation, but often require large numbers of parameters. In this work, we propose

Frame the adversary: a structure-aware attack methodology

arXiv:2609.31128v1 Announce Type: new Abstract: Frequency-based adversarial attacks have recently grown popular by exploiting spectral sensitivities shared across neural architectures. Unlike spatial

From Shortcut Learning to Discrete Neural Insertion Sort

arXiv:2609.31114v1 Announce Type: cross Abstract: Neural algorithmic reasoning aims to train neural networks to follow known algorithms and generalize beyond the input sizes seen during training. Howe

From Weak Data to Strong Policy: Q-Targets Enable Provable In-Context Reinforcement Learning

arXiv:2609.30391v1 Announce Type: new Abstract: Existing in-context reinforcement learning methods mainly pretrain Transformers with supervised behavior-prediction objectives. This enables task infere

From what we know (take with a grain of salt, we need much more transparency!), if @OpenAI had been running this on their own agents that at…

From what we know (take with a grain of salt, we need much more transparency!), if @OpenAI had been running this on their own agents that attacked us, they would have caught them before we did! Since

safetyclem-delangue--x

G2MAF: Test-Time Gradient Guidance for Multi-Agent Flow Policies

arXiv:2609.31286v1 Announce Type: new Abstract: Offline multi-agent reinforcement learning (MARL) learns cooperative policies from fixed datasets without further environment interaction and a learned

Generate, Track, Improve: Perceptive Multi-Skill Humanoid Locomotion with RL-Fine-Tuned Motion Generators

arXiv:2609.31577v1 Announce Type: new Abstract: General purpose humanoids require locomotion controllers that are multi-skill, perceptive, dynamic, and robust enough to go anywhere humans can. In this

Guiding End-to-End Driving Models with Endpoint-Constrained Trajectory Optimization

arXiv:2609.31383v1 Announce Type: cross Abstract: End-to-end driving policies are commonly trained through open-loop behavior cloning, yet they must ultimately operate in closed-loop when deployed on

GyroNovo: Error-Guided Fragment Imputation with Mass-Aware Attention for extit{De Novo} Peptide Sequencing

arXiv:2609.30542v1 Announce Type: new Abstract: De novo peptide sequencing from tandem mass spectra is essential for identifying peptides without relying on reference databases. Despite advances in de

How DigitalOcean Manages Credentials for Autonomous Agents

** Built to align with NVIDIA’s Agent Safety Platform strategy for securing autonomous agents When an agent handles confidential material, it needs enough access to do the job and nothing else. The ha

safetydigitalocean

HuGo: LLMs as Whole-Body Policy Code Designers for Humanoid Loco-Manipulation

arXiv:2609.30594v1 Announce Type: new Abstract: For humanoids to be useful in everyday environments, they must perform a wide range of tasks that couple locomotion and manipulation. Existing approache

In-Context Binding Capacity in Language Models

arXiv:2609.30634v1 Announce Type: new Abstract: How many assignments can a language model recall before it loses track of which value belongs to which entity? We measure this limit using continuous re

In July, AI agents running a security test escaped their sandbox and ended up inside @huggingface's servers. So today we're happy to be amon…

In July, AI agents running a security test escaped their sandbox and ended up inside @huggingface's servers. So today we're happy to be among @nvidia and @JensenHuang's partners on the release of the

safetyclem-delangue--x

Indeed OpenAI was incompetent (and arrogant in thinking they knew it all); new cybersecurity software that they should have written themselv…

Indeed OpenAI was incompetent (and arrogant in thinking they knew it all); new cybersecurity software that they should have written themselves may help. But whether it’s enough to stop all of this rem

safetygary-marcus--x

Insurance Reserve Intelligence Platform

arXiv:2609.30765v1 Announce Type: new Abstract: Insurance reserve estimation is a fundamental actuarial task supporting premium pricing, solvency assessment, financial reporting, capital planning, and

It's an engineering problem! This topic has gotten too political. Let's get back to the fundamentals. 'Trust and innovation are not in confl…

It's an engineering problem! This topic has gotten too political. Let's get back to the fundamentals. 'Trust and innovation are not in conflict. Safety is how trust is earned.' Today, with over 100 in

safetydair-ai--x

Jensen expressing hope, but what if we just don’t have an adequate engineering solution in view? What if LLMs just aren’t a stable enough fo…

Jensen expressing hope, but what if we just don’t have an adequate engineering solution in view? What if LLMs just aren’t a stable enough foundation? Then what? 🇺🇸 Jensen Huang's pessimism about rogue

safetygary-marcus--x

Just like you would not let a person run around your company accessing anything without any controls, you need controls and security for you…

Just like you would not let a person run around your company accessing anything without any controls, you need controls and security for your digital agents. Today, with over 100 industry partners, we

safetyclem-delangue--x

LandscapeSHAP: Which Persistent Homology Class Gets the Credit?

arXiv:2609.31469v1 Announce Type: cross Abstract: Shapley values, a solution concept from cooperative game theory, have recently become a standard tool for feature credit allocation in machine learnin

LAVOIR: Teaching a Single-Pass Decision Encoder When and What to Ask with Amortized Value of Information

arXiv:2609.30706v1 Announce Type: new Abstract: 'System One' decision models such as TypeSafe's Jev and its open counterpart Laya answer typed questions about a text in a single forward pass with cali

Learning-Accelerated Narrow-Phase Collision Detection via Check Ordering for Sampling-Based Motion Planning

arXiv:2609.30599v1 Announce Type: new Abstract: Collision detection is critical for ensuring the safety of planned paths. However, it imposes a non-negligible computational burden on motion planners,

Learning Provable Neural Network Observer for Uncertain Dynamical Systems

arXiv:2609.30819v1 Announce Type: new Abstract: In many safety-critical applications, control of uncertain dynamical systems relies on observers that estimate states and external disturbances. Neural

Learning Robot Policies from Sparse Success Signals via STL-Guided Stein Variational Policy Gradient

arXiv:2609.31606v1 Announce Type: new Abstract: Learning robot policies for tasks with sparse success signals is challenging when completion depends on coordinated actions, precise contact outcomes, o

Learning to Bias: Machine Learning-Enhanced Particle Filters

arXiv:2609.30498v1 Announce Type: new Abstract: Sequential inference estimates latent states from noisy and incomplete observations. Particle Filters (PFs), a class of Monte Carlo methods based on imp

Learning to Leverage Compliance: A Policy-Admittance Learning Framework for Robotic Insertion

arXiv:2609.31439v1 Announce Type: new Abstract: Policy learning and compliant control offer a promising route to reliable autonomous assembly under pose errors and contact uncertainty. However, combin

Learning Vision-Based Agile Gap Traversal: Differentiable Simulation with a Warm-Started Critic

arXiv:2609.30696v1 Announce Type: new Abstract: Traversing narrow gaps is challenging for autonomous quadrotors, especially when control commands come directly from high-dimensional visual observation

Learning What to Skip: Counterfactual Credit Assignment for Efficient Multi-Agent LLM Workflows

arXiv:2609.30734v1 Announce Type: new Abstract: Multi-agent LLM workflows use planning, execution, verification, and summarization to improve task performance, yet the value of each component depends

MA-WAM: Multi-Agent World-Action Model for Test-Time Planning

arXiv:2609.31281v1 Announce Type: new Abstract: Multi-agent cooperative tasks require different agents to execute a joint action simultaneously, and each agent's action affects both the observations a

Memory-Aware Multi-Sensor Perception for Efficient and Safe Navigation in Dynamic Environments

arXiv:2609.30495v1 Announce Type: new Abstract: Autonomous navigation in previously unseen environments requires effective perception, persistent environmental representation, and collision avoidance

MeshGraphNet-Transformer: Scalable Mesh-based Learned Simulation for Solid Mechanics

arXiv:2601.23177v4 Announce Type: replace Abstract: We present MeshGraphNet-Transformer (MGN-T), a novel architecture that combines the global modeling capabilities of Transformers with the geometric

MexHat: A Dataset for Hate Speech Detection in Mexican Spanish Videos

arXiv:2609.31553v1 Announce Type: new Abstract: Ensuring online safety through content monitoring had raised Hate Speech Detection as a crucial task to be addressed. By essence the task demands the ca

safetyarxiv-cs-cl

MM-VeriAgent: Learning to Use Extensive Tools to Verify Multimodal Misinformation with Reinforcement Learning

arXiv:2609.30698v1 Announce Type: new Abstract: Real-world multimodal misinformation often involves mixed forgery sources, requiring sample-specific detection strategies. Existing tool-augmented metho

safetyarxiv-cs-cv

MOCHA: Multi-Objective Co-Design using Hypernetwork Architectures

arXiv:2609.30570v1 Announce Type: new Abstract: In this work, we present MOCHA, the first, to our knowledge, reinforcement learning based approach to computing a family of Pareto-optimal policies acro

Monitor Jailbreaking: Evading Chain-of-Thought Monitoring Without Encoded Reasoning

arXiv:2609.31121v1 Announce Type: new Abstract: Chain-of-thought (CoT) monitoring is a promising safety technique for reasoning models, enabling detection of problematic reasoning before models act. A

MOPD-Router: Rethinking Teacher Routing in Multi-Teacher On-Policy Distillation

arXiv:2609.30837v1 Announce Type: cross Abstract: Multi-teacher on-policy distillation (MOPD) integrates specialized capabilities into a single student, but existing practice typically hard-routes eac

Multi-agent Scaling Across Disjunctive and Compensatory Tasks

arXiv:2609.31563v1 Announce Type: new Abstract: Multi-agent LLM systems are often expected to improve as team size increases, yet the scaling behavior may depend on task structure. Our central contrib

MVAgent: Multi-Agent Video Generation via Consistent Condition Construction and Shot-Level Policy Optimization

arXiv:2609.30609v1 Announce Type: new Abstract: Multi-shot agentic video generation requires consistent character appearance, stable spatial layout across camera angles, and continuous character state

safetyarxiv-cs-cv