safety

Towards AI epidemiology: a measurement standardisation framework for prospective risk detection

arXiv:2512.15783v3 Announce Type: replace Abstract: This paper proposes a measurement standardisation framework that compresses expert-AI interactions into structured, comparable fields for prospectiv

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Towards Healthy Evolution: Exploring the Role and Mechanisms of Human-Agent Interaction in Self-Evolving Systems

DGX agent

arXiv:2606.06114v1 Announce Type: new Abstract: Self-evolving agents improve through continual self-play and self-generated learning signals, but autonomous evolution can also cause capability degrada

safetyarxiv-cs-ai

6 Jun 2026

Towards World Models in Biomedical Research

arXiv:2606.05925v1 Announce Type: new Abstract: A central goal of biomedicine is to understand, predict and ultimately control the dynamic mechanisms by which biological systems respond to perturbatio

UniVoice: A Unified Model for Speech and Singing Voice Generation

arXiv:2606.05852v1 Announce Type: cross Abstract: Text-to-speech (TTS) and singing voice synthesis (SVS) both aim to generate human vocal audio from symbolic inputs, but they impose different requirem

Unsupervised Pattern Analysis in Japanese Veterinary Toxicology: A Regulatory-Compliant Framework for Cross-Species Risk Assessment

arXiv:2606.06207v1 Announce Type: new Abstract: Veterinary pharmacovigilance systems are essential for monitoring adverse drug events (ADEs), yet existing approaches often fail to capture region-speci

We live at a delicate, tragic moment in history, and greed and desperation is probably about to make it much worse.

Gary Marcus expresses concern about contemporary global instability, suggesting that human greed and desperation pose significant risks to an already precarious historical moment. The post implies tha

Where Should Knowledge Enter? A Layered Framework for Knowledge Infusion in Multimodal Iterative Generative Mo

arXiv:2606.06356v1 Announce Type: new Abstract: Multimodal generative models produce fluent outputs but remain unreliable when generation must respect structured, domain-specific, or safety-critical k

Whether SpaceX/ Xai is making money on the deals with Google and Anthropic or losing money, they are waving the towel on winning the frontie…

Whether SpaceX/ Xai is making money on the deals with Google and Anthropic or losing money, they are waving the towel on winning the frontier model race— by arming their competitors rather than themse

White House AI advisor Sriram Krishnan says he will leave his role at the end of June; sources: Krishnan plans to start a pro-Trump AI policy institution (Leo Schwartz/The Information)

Leo Schwartz / The Information: White House AI advisor Sriram Krishnan says he will leave his role at the end of June; sources: Krishnan plans to start a pro-Trump AI policy institution — Sriram Krish

safetytechmeme

wild thought: will US invest in Anthropic, the alleged supply chain risk? weird if they do. but if they don’t (but do invest in OpenAI) the …

wild thought: will US invest in Anthropic, the alleged supply chain risk? weird if they do. but if they don’t (but do invest in OpenAI) the US government may immensely and immediately increase Anthrop

Willing but Unable: Separating Refusal from Capability in Code LLMs via Abliteration

arXiv:2606.05396v1 Announce Type: cross Abstract: Producing a labeled vulnerable code at scale is a recurring obstacle for learning-based vulnerability detection: mined corpora carry substantial label

Your GFlowNet Secretly Learns an Optimal Transport Plan

arXiv:2606.06272v1 Announce Type: cross Abstract: Generative Flow Networks (GFlowNets) are a framework for sampling structured objects via stochastic trajectories in a directed graph. In this work, we

Zero knowledge verification for frontier AI training is possible

arXiv:2606.05433v1 Announce Type: new Abstract: Frontier AI governance frameworks increasingly use cumulative training compute as the primary criterion for designating high-impact models, but enforcem

A Komi-Yazva--Russian Parallel Corpus and Evaluation Protocol for Zero- and Few-Shot LLM Translation

arXiv:2606.06420v1 Announce Type: new Abstract: We present the first Komi-Yazva--Russian parallel corpus together with an explicit evaluation protocol for studying LLM translation in an endangered, ex

A Systematic Analysis of Biases in Large Language Models

arXiv:2512.15792v3 Announce Type: replace-cross Abstract: Large language models (LLMs) have rapidly become indispensable tools for acquiring information and supporting human decision-making. However,

Absofuckinglutely called it. Nationalized stakes are just a bailout by a different name. Here we are seventeen months later, and the fleece …

Absofuckinglutely called it. Nationalized stakes are just a bailout by a different name. Here we are seventeen months later, and the fleece the taxpayer game is on. The countdown until we are told tha

ACE-SQL: Adaptive Co-Optimization via Empirical Credit Assignment for Text-to-SQL

arXiv:2606.05906v1 Announce Type: new Abstract: Text-to-SQL maps natural language questions to executable SQL queries. Modern databases often contain large and complex schemas, making schema linking a

Adversarial Attacks Already Tell the Answer: Directional Bias-Guided Test-time Defense for Vision-Language Models

arXiv:2606.06186v1 Announce Type: new Abstract: Vision-Language Models (VLMs), such as CLIP, have shown strong zero-shot generalization but remain highly vulnerable to adversarial perturbations, posin

Alignment Risks from Capability-Seeking RL Training

arXiv:2602.12124v2 Announce Type: replace-cross Abstract: While most AI alignment research focuses on preventing models from generating explicitly harmful content, a more subtle risk arises from capab

“America won’t win the AI race if we beat China but end up with a CCP-style social credit system in the U.S. — and that is the danger as the…

“America won’t win the AI race if we beat China but end up with a CCP-style social credit system in the U.S. — and that is the danger as the government becomes more deeply involved in AI development a

Analysis of the Neglect-Zero Effect in Large Language Models

arXiv:2606.05864v1 Announce Type: new Abstract: We investigate the extent to which the language processing of LLMs resembles human cognitive processes, focusing on a human cognitive bias called the ex

Attitude-Aided Linear Calibration of Triaxial Accelerometers

arXiv:2606.06308v1 Announce Type: new Abstract: Triaxial MEMS accelerometers are widely used for inertial sensing, navigation, and sensor fusion, but existing calibration methods often rely on costly

Auditing Demonstration Curation Metrics: Action-Only Scorers Fail on the Structural Defects That Degrade Imitation Policies

arXiv:2606.05588v1 Announce Type: new Abstract: Imitation-learning policies inherit the quality of the demonstrations they are trained on, and a growing set of curation metrics promise to score and fi

Beyond Alignment: Value Diversity as a Collective Property in Multicultural Agent Systems

arXiv:2606.05985v1 Announce Type: new Abstract: Multicultural multi-agent systems are increasingly deployed in globally diverse settings, where different agents are grounded in different cultural back

Beyond Imitation: Reinforcement Learning-Based Sim-Real Co-Training for VLA Models

arXiv:2602.12628v4 Announce Type: replace Abstract: Simulation offers a scalable and low-cost way to enrich vision-language-action (VLA) training, reducing reliance on expensive real-robot demonstrati

Beyond tokens: a unified framework for latent communication in LLM-based multi-agent systems

arXiv:2606.05711v1 Announce Type: new Abstract: Multi-agent systems built on large language models (LLMs) have become a prevailing paradigm for tackling complex reasoning, planning, and tool-use tasks

CHASE: Adversarial Red-Blue Teaming for Improving LLM Safety using Reinforcement Learning

arXiv:2606.05523v1 Announce Type: new Abstract: Despite advances in safety alignment, prompt-rewriting attacks such as persona modulation, fictional framing and persuasion-based reformulation, can byp

'Chi nas dal soch el sent de legn' -- Auditing Text Corpora for Lombard

arXiv:2606.06349v1 Announce Type: new Abstract: Several of the world's languages are still under-resourced in terms of Natural Language Processing (NLP) tools. This is mostly due to the lack of high-q

Cosine Misleads: Auxiliary Losses Reshape Vision Language Models, Not Their Latents

arXiv:2606.05753v1 Announce Type: new Abstract: Latent visual reasoning (LVR) inserts supervised latent tokens between perception and answer generation in vision-language models (VLMs). The field uses

DexFuture: Hierarchical Future-State Visuomotor Targeting for Bimanual Dexterous Tool Use

arXiv:2606.05699v1 Announce Type: new Abstract: Bimanual dexterous tool use remains challenging for robots due to high-dimensional hand configurations and complex hand-tool-object dynamics and contact

Discrete-WAM: Unified Discrete Vision-Action Token Editing for World-Policy Learning

arXiv:2606.05645v1 Announce Type: new Abstract: Autonomous driving requires reasoning about how ego actions shape the evolution of the surrounding world. However, most end-to-end methods rely on direc

Disentangled Fine-Grained Prototype Learning for Incomplete Image-Tabular Classification

arXiv:2606.05455v1 Announce Type: new Abstract: The missing-modality problem poses a significant challenge in image-tabular multimodal learning across a wide range of multimedia applications, includin

Do Models Share Safety Representations? Cross-Model Steering for Safe Visual Generation

arXiv:2606.05290v1 Announce Type: new Abstract: Recent progress in generative modeling has made safety control a central challenge, yet existing approaches remain largely model-specific, requiring ret

Drishti AI-Event Guardian: An Intelligent Real-Time Crowd Monitoring and Emergency Response System for Mass Gathering Events

arXiv:2606.05185v1 Announce Type: cross Abstract: Mass gathering events are associated with critical safety incidents caused by insufficient crowd monitoring and inadequate emergency response coordina

EgoHumanoid: Unlocking In-the-Wild Loco-Manipulation with Robot-Free Egocentric Demonstration

arXiv:2602.10106v2 Announce Type: replace Abstract: Human demonstrations offer rich environmental diversity and scale naturally, making them an appealing alternative to robot teleoperation. While this

EMBER: Efficient Memory via Budgeted Evidence Retention for Long-Horizon Agents

arXiv:2606.05894v1 Announce Type: new Abstract: Long-horizon agents can archive large histories, but future answers still incur retrieval, rereading, and context costs. When retained memory misses ans

Ensuring Interaction Safety in Multitask Exoskeleton Control: A Simulation-Trained Variable Impedance Framework

arXiv:2606.06370v1 Announce Type: new Abstract: Wearable exoskeletons can augment human phys ical capabilities during complex activities. However, ensuring adaptation across diverse tasks while guaran

Epistemic Injustice in Language Models: An Audit of Pretraining Filters and Guardrails

arXiv:2606.05936v1 Announce Type: new Abstract: Modern language models rely on pretraining filters to remove undesirable content from training corpora and inference-time guardrails to suppress undesir

EVE: A Generator-Verifier System for Generative Policies

arXiv:2512.21430v2 Announce Type: replace Abstract: Visuomotor policies based on generative such as diffusion and flow-matching have shown strong performance for robotics applications but degrade unde

Finally, a commencement speaker who calls bullshit. Great oped by @mollyjongfast on “billionaire brain”, and why young people have a right t…

Finally, a commencement speaker who calls bullshit. Great oped by @mollyjongfast on “billionaire brain”, and why young people have a right to boo what the AI industry has become. I wrote about the com

Flow-based Policy Adaptation without Policy Updates

arXiv:2606.06461v1 Announce Type: new Abstract: Leveraging prior knowledge from pretrained policies, foundation models, or human operators offers an efficient alternative to learning robot skills from

FlowPRO: Reward-Free Reinforced Fine-Tuning of Flow-Matching VLAs via Proximalized Preference Optimization

arXiv:2606.05468v1 Announce Type: new Abstract: Post-training Vision-Language-Action (VLA) models into policies that can be reliably deployed on real robots remains a major bottleneck. SFT and DAgger

Forgive or forget: Understanding the context of hate in audio retrieval systems

arXiv:2606.05857v1 Announce Type: new Abstract: Handling toxic retrieval in text-to-audio systems is challenging due to contextual dependencies. Existing strategies (e.g., rephrasing, summarization) r

Funny how so many people read Anthropic is calling for a pause when they did NOT actually call for a pause. Read what they said, carefully. …

Funny how so many people read Anthropic is calling for a pause when they did NOT actually call for a pause. Read what they said, carefully. They want it both ways. They *don’t* actually want a pause -

Geometry-Aware Dataset Condensation for Diffusion Model Training

arXiv:2606.05883v1 Announce Type: new Abstract: Dataset condensation aims to construct compact datasets from real data via synthesis or selection. However, existing approaches are ill-suited for diffu

GLASS: GRPO-Trained LoRA for Acoustic Style Steering in Zero-Shot Text-to-Speech

arXiv:2606.05889v1 Announce Type: cross Abstract: We propose GLASS, a framework for composable acoustic style control in zero-shot autoregressive text-to-speech (TTS) that learns controls from post-ge

Grounded but Misleading: Evaluating Semantic Alignment in AI-Generated Security Explanations

arXiv:2602.05056v2 Announce Type: replace-cross Abstract: Online scams increasingly leverage fluent and context-aware social engineering strategies, creating growing demand for AI systems that explain

HANDOFF: Humanoid Agentic Task-Space Whole-Body Control via Distilled Complementary Teachers

arXiv:2606.06493v1 Announce Type: new Abstract: For a humanoid robot to be deployed in the real world, the choice of command space (i.e., the interface between task planning and whole-body control) is