Search: “safety”

LLM Probability Concentration: How Alignment Shrinks the Generative Horizon

arXiv:2506.17871v4 Announce Type: replace-cross Abstract: Despite their impressive capabilities, aligned large language models (LLMs) often generate outputs that lack diversity. What drives this consi

15 Sept 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

PIDS-Bench: Evaluating Prompt-Injection Detectors Under Over-Defense, Obfuscation, and Distribution Shift

DGX agent

arXiv:2609.15017v1 Announce Type: cross Abstract: Prompt-injection detectors are typically evaluated using aggregate F1 on in-distribution test data, which offers limited insight into behavior under d

model-releasesarxiv-cs-ai

15 Sept 2026

The Misery of Mechanistic Interpretability: A Formal Perspective

arXiv:2609.15533v1 Announce Type: cross Abstract: Mechanistic interpretability has become the dominant lens for understanding frontier language models, as their inner workings are complex and inherent

15 Sept 2026

Dan Selsam is a current OpenAI capabilities researcher. (since 2022) He was my boss for a while. He doesn't have a twitter account but has m…

model-releasessonya-huang--x

Dan Selsam is a current OpenAI capabilities researcher. (since 2022) He was my boss for a while. He doesn't have a twitter account but has made this public statement of his views on AI risk and sent i

14 Sept 2026

EgoMaize: A First-Person Maize Instance Segmentation Benchmark under Severe Field Occlusion

arXiv:2609.12350v1 Announce Type: new Abstract: Close-range first-person field images are important for mobile maize phenotyping because many plant-level traits depend on in-canopy structures that are

14 Sept 2026

ORQA: An Occupation-Realistic Question and Answer Framework for LLM Professional Knowledge

arXiv:2609.12366v1 Announce Type: new Abstract: We present ORQA, a method for testing occupation-level knowledge in large language models. Prior methods either map abstract LLM skills to occupations v

14 Sept 2026

Wild 24 hours for AI and lots of different proposals have been made. TLDR; the only tangible new fact is that OpenAI and Anthropic are goi…

model-releaseselon-musk--x

Wild 24 hours for AI and lots of different proposals have been made. TLDR; the only *tangible* new fact is that OpenAI and Anthropic are going to have embedded 3rd party evaluators from unknown organi

13 Sept 2026

Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation

arXiv:2609.11115v1 Announce Type: new Abstract: Benchmark researchers and developers of large language models (LLMs) and other AI systems need to find relevant evaluations, locate their benchmark data

12 Sept 2026

AgentAudit: An Open, Extensible Framework for Full-Lifecycle Trust Evaluation of AI Agents

arXiv:2609.09875v1 Announce Type: new Abstract: Existing evaluation frameworks mostly assess only one part of AI agents, such as task completion (AgentBench) or security robustness (AgentDojo, ASB), r

AI Alarm Hype Cycles - A brief history GPT-2: OpenAI withholds weights due to misuse/misinformation concerns GPT-3: No open release, Dario A…

model-releasesclem-delangue--x

AI Alarm Hype Cycles - A brief history GPT-2: OpenAI withholds weights due to misuse/misinformation concerns GPT-3: No open release, Dario Amodei and other GPT-3 team members depart OpenAI to found An

Certifying Lower Bounds for Risk-Sensitive Reinforcement Learning under Adversarial State Perturbations

arXiv:2609.10866v1 Announce Type: new Abstract: Reinforcement learning (RL) agents deployed in real-world environments are often vulnerable to adversarial perturbations in state observations, creating

Safe to Stop? Risk-Constrained Stopping for Sequential Clinical Diagnosis Agents

arXiv:2609.09678v1 Announce Type: new Abstract: Clinical diagnosis agents must decide not only what test to request next, but also when to diagnose or defer. Existing agent benchmarks largely evaluate

Towards Reliable Medical LLMs: Benchmarking and Enhancing Confidence Estimation of Large Language Models in Medical Consultation

arXiv:2601.15645v2 Announce Type: replace Abstract: Large-scale language models (LLMs) often offer clinical judgments based on incomplete information, increasing the risk of misdiagnosis. Existing stu

Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning

arXiv:2609.10142v1 Announce Type: new Abstract: Large language models remain fragile against malicious fine-tuning, motivating training-time defenses against harmful persona drift. Preventative Steeri

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models

arXiv:2603.20957v4 Announce Type: replace-cross Abstract: Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data. They further rely

EmoMed: An Emotionally-Aware Agent for Multimodal Medical Support with Real-Time Information Retrieval

arXiv:2609.07194v1 Announce Type: new Abstract: We present EmoMed - a multimodal medical consultation agent that adapts its responses based on users' emotional states while maintaining clinical accura

HaWMPO: Hallucination-Aware World Model-based Policy Optimization for Generalist Robot Policy

model-releasesarxiv-cs-ro

arXiv:2609.09941v1 Announce Type: new Abstract: Generalist robot policies have demonstrated strong generalization across robotic manipulation tasks, yet their success rates remain limited in com- plex

PiMRef: Deducing Ever-evolving Spear-phishing Emails with Knowledge Base Invariants

arXiv:2507.15393v2 Announce Type: replace-cross Abstract: Phishing email is a critical step in the cybercrime kill chain due to the high reachability of victims' email accounts and the low cost of lau

Quality Metrics for LLM-Generated Asset Administration Shells: A Perturbation-Based Evaluation Approach

arXiv:2609.07290v1 Announce Type: cross Abstract: The rapid digital transformation of manufacturing, often referred to as Industry 4.0, relies on seamless interoperability between physical and softwar

SchemeArena: Factorized Stress Testing of Scheming in LLM Agents

arXiv:2609.08126v1 Announce Type: new Abstract: We study scheming in LLM agents, in which agents covertly pursue misaligned goals. Our focus is to understand how scheming arises from the interaction o

Skynet: Workflow-Level Anomaly Detection for Agentic AI via Semantic and Structural Modeling

arXiv:2609.06835v1 Announce Type: cross Abstract: Agentic AI systems execute complex tasks through long-horizon workflows of planning, tool use, and multi-agent coordination. Task failures in these sy

local-aiarxiv-cs-ai

The Unreliable Progress Bar: Can LLM Agents Reliably Report Task Progress Throughout Execution?

arXiv:2609.08589v1 Announce Type: cross Abstract: Recent large language models can emit task-progress signals that agent frameworks use to decide whether a task should continue or stop, yet whether a

VANTAGE-Bench: Evaluating the Infrastructure AI Gap in Vision-Language Models

arXiv:2609.09396v1 Announce Type: new Abstract: As Vision-Language Models (VLMs) advance toward physical deployment, the focus has remained on action-oriented Embodied AI evaluated on subject-centric

Your Agent Says Yes: Interpreting Adversarial Market Behavior Beyond Individual Transactions

arXiv:2609.07675v1 Announce Type: cross Abstract: Transaction-local controls answer whether one financial request may proceed, but market behavior can be distributed across messages, agents, assets, a

local-aiarxiv-cs-ai

Accuracy is Not Enough: A Divergence-Based Approach to Evaluate Fidelity Loss in Quantized LLMs

arXiv:2609.07664v1 Announce Type: new Abstract: Deployment of Large Language Models (LLMs) on memory-constrained edge devices relies heavily on aggressive post-training quantization. However, evaluati

BlueprintAgent: Constraint-Triggered Targeted Revisits for Simulation-Ready Generation from Scanned Structural Blueprints

arXiv:2609.07362v1 Announce Type: cross Abstract: Converting in-service reinforced-concrete (RC) building blueprints into simulation-ready models---structured frame representations that support determ

local-aiarxiv-cs-cv

ChatBEV: Empowering Traffic Scene Understanding and Simulation via Vision-Language Model

arXiv:2503.13938v3 Announce Type: replace Abstract: Comprehensive traffic scene understanding is a foundational capability for Intelligent Transportation Systems (ITS) underpinning applications such a

CoRL: Co-Evolutionary Reinforcement Learning for Adaptive Indirect Prompt-Injection Attacks and Defenses

arXiv:2609.07529v1 Announce Type: new Abstract: Tool-augmented language agents are vulnerable to indirect prompt injection (IPI). Unlike direct prompt injection, IPI hides adversarial instructions in

Event-Based De-Snowing for Autonomous Driving

arXiv:2507.20901v2 Announce Type: replace Abstract: Adverse weather conditions, particularly heavy snowfall, pose significant challenges to both human drivers and autonomous vehicles. Traditional imag

From LLM-Generated Specifications to Learned Quadruped Locomotion

model-releasesarxiv-cs-ro

arXiv:2609.07111v1 Announce Type: new Abstract: Quadruped robot locomotion policies are often trained using reinforcement learning, which in turn relies heavily on hand-crafted reward functions. Desig

Information-Guided Safe Reinforcement Learning for Autonomous Gas Source Localization using sUAS

arXiv:2609.05569v1 Announce Type: new Abstract: The autonomous localization of fugitive gas emissions using small Unmanned Aircraft Systems (sUAS) constitutes a fundamentally ill-posed inverse problem

local-aiarxiv-cs-ro

Novel Methods for Catheter and Guidewire Segmentation in X-ray Fluoroscopy under a Federated Learning Setting

arXiv:2609.06876v1 Announce Type: new Abstract: Endovascular procedures rely on real-time manipulation of thin instruments, catheters and guidewires, under X-ray fluoroscopy guidance, where accurate v

SAM3-O2D2: Zero-Shot Object Out-of-Distribution Detection by Object Class Prompting of the SAM3-Image Model

arXiv:2609.08281v1 Announce Type: new Abstract: Object detectors have shown remarkable performance in various fields, among these medical imaging, surveillance, and autonomous driving. However, they a

local-aiarxiv-cs-cv

This is a really neat benchmark. It's super interesting that Astra starts strong and holds the lead for up to 19 hours, but Fable 5.1 catche…

This is a really neat benchmark. It's super interesting that Astra starts strong and holds the lead for up to 19 hours, but Fable 5.1 catches up in the final hours. Not entirely clear why. Different m

model-releasesdair-ai--x

Time-Aware Assistive Navigation

arXiv:2609.05596v1 Announce Type: new Abstract: Can interactive vision-and-language agents learn not just what to say but also extbf{extit{when}} to say it? Current language models rarely plan over wh

b10857

model-releasesllama-cpp-releases

Fix Vulkan-Hpp handle usage on 32-bit targets. (#22892) On 32-bit platforms, Vulkan non-dispatchable handles such as VkBuffer are represented as uint64_t, and Vulkan-Hpp disables implicit conversions

8 Sept 2026

I am pleased to see that OpenAI’s new model is roughly on par with Gemini Flash and Opus 4.8 on prompt injection risk. Nice work! Evaluating…

model-releasesboris-cherny--x

I am pleased to see that OpenAI’s new model is roughly on par with Gemini Flash and Opus 4.8 on prompt injection risk. Nice work! Evaluating and naming other labs turns out to be a great way to encour

8 Sept 2026

Hardware

Introducing CUDA Rust: Two Tracks for Writing GPU Kernels

NVIDIA introduced two Rust projects for GPU development in September 2026: `cuda‑oxide`, a nightly‑only backend that compiles SIMT‑style kernels to PTX via LLVM and uses DisjointSlice and launch contr

hardwarenvidia-developer

8 Sept 2026

After over a year of my nights and weekends, the Jenny app is done!

model-releasesr-localllama

Hi all! I just wanna say that I am tired lol. Yes, it's another harness, but I spent a lot of time and effort and have forsaken my hobbies to build the Jenny (like XJ-9) app. Jenny is a free, MIT lice

7 Sept 2026

Knowing What Not to Answer: Selective Non-Compliance in Vision-Language Models

arXiv:2609.04720v1 Announce Type: cross Abstract: Vision-language models (VLMs) are expected to respond helpfully to appropriate requests while withholding compliance with requests that are incorrect,

7 Sept 2026

Beyond Majority Vote: Multi-Perspective Adjudication for Medical Hallucination Detection

arXiv:2609.03953v1 Announce Type: new Abstract: Understanding the frequency of factual errors in chatbot-generated text and evaluating systems that detect these errors is critical for determining chat

Beyond Reproducibility: Towards Security-Aware Evaluation of Research Artifacts

arXiv:2605.06508v2 Announce Type: replace-cross Abstract: Research artifacts are widely shared to support reproducibility, and artifact evaluation (AE) has become common at many leading conferences. H

BharatGather: A Culturally-Informed Benchmark Dataset for Misinformation and Fake News Detection in Indian Public Events

arXiv:2609.02895v1 Announce Type: new Abstract: Large-scale public events, such as religious festivals, political rallies, and cultural gatherings, are increasingly vulnerable to the rapid disseminati

Hardware

Building a Memory-Driven Agent with NVIDIA NemoClaw

NVIDIA NemoClaw delivers a memory‑driven Chief of Staff agent that constructs a structured self‑model (people, projects, priorities) stored in Markdown pages, with an SQLite ledger recording obligatio

hardwarenvidia-developer

CASCADE: A Component Ablation and Corpus Audit of a Layered Local Defense for MCP-Based Systems

arXiv:2604.17125v2 Announce Type: replace-cross Abstract: The Model Context Protocol (MCP) widens the prompt injection attack surface of large language model applications to tool descriptions, paramet

Inferred Generative-Process Diversity Predicts Correlated Failure Across Language Models

arXiv:2609.03422v1 Announce Type: new Abstract: Diversity is a widely observed factor in the resilient function of collective systems, yet the type of diversity that matters depends on the properties

Privacy Leakage in Federated Learning: Gradient-Based Client Identity Inference and Defenses for Inertial Sensing in Vehicular Edge Networks

arXiv:2609.02971v1 Announce Type: cross Abstract: As vehicular networks move toward 5G/6G edge intelligence, federated learning (FL) is widely promoted as a privacy-preserving way for vehicles and inf

CoMerge: Conflict-Driven Preference Optimization for Multi-Task Model Merging

arXiv:2609.02273v1 Announce Type: new Abstract: Model merging provides an efficient paradigm for constructing multi-task large language models (LLMs) without full model retraining, yet it remains chal