safety

A Structural Theory of Position Bias in Transformers

arXiv:2602.16837v2 Announce Type: replace Abstract: Transformer models systematically favor certain token positions, yet the architectural origins of this position bias remain poorly understood. This

Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

DGX agent

arXiv:2605.28664v1 Announce Type: cross Abstract: Safety detection models require examples of HHH (Helpful, Harmless, Honest)-violating outputs for robust generalization, however such examples are sca

safetyarxiv-cs-cl

28 May 2026

AdaMemento: Adaptive Memory-Assisted Policy Optimization for Reinforcement Learning

arXiv:2410.04498v2 Announce Type: replace Abstract: In sparse reward scenarios of reinforcement learning (RL), the memory mechanism provides promising shortcuts to policy optimization by reflecting on

ADWIN: Adaptive Windows for Horizon-Aware On-Policy Distillation

arXiv:2605.28396v1 Announce Type: cross Abstract: On-policy distillation (OPD) transfers reasoning behavior by training a student on teacher feedback along student-generated trajectories, but standard

Affective Music Recommendation: A Rollout-Based World Model for Offline Preference Optimization

arXiv:2605.28810v1 Announce Type: new Abstract: Functional music applications, from consumer focus and sleep aids to clinical interventions, share a distinctive recommendation problem: success is defi

AgensFlow: A Coordination-Policy Substrate for Multi-Agent Systems

arXiv:2605.27466v1 Announce Type: cross Abstract: Multi-agent systems built on large language models (LLMs) require many coordination choices that are difficult to fix a priori: which skill protocol t

Agent Explorative Policy Optimization for Multimodal Agentic Reasoning

arXiv:2605.28774v1 Announce Type: new Abstract: Vision-language models with extended reasoning succeed on complex problems, but many real-world problems require external tools that internal reasoning

Agents that Matter: Optimizing Multi-Agent LLMs via Removal-Based Attribution

arXiv:2605.27621v1 Announce Type: cross Abstract: As multi-agent systems (MAS) become increasingly complex, identifying the contributions of individual agents is critical for system optimization. Howe

AI systems may soon help run economies, infrastructure, and military operations. But these systems are not reliably loyal or secure. An adve…

AI systems may soon help run economies, infrastructure, and military operations. But these systems are not reliably loyal or secure. An adversary can make an AI work against its own operator. In our n

safetydan-hendrycks--x

AI, Take the Wheel: What Drives Delegation and Trust in Human-Computer Cooperative Question Answering?

arXiv:2605.28255v1 Announce Type: new Abstract: AI systems are fallible, and humans can make mistakes in deciding whether to trust AI over their own judgment. Thus, improving human-AI collaboration re

all these decades i have been using calculators, and it never occurred to me that they might be … conscious!

Gary Marcus raises a philosophical question about whether calculators, despite their computational capabilities, might possess consciousness. The post likely explores the disconnect between functional

Almost everyone is building agent harness systems the wrong way. The default move: pick LangChain or LangGraph or the OpenAI Agents SDK, acc…

Almost everyone is building agent harness systems the wrong way. The default move: pick LangChain or LangGraph or the OpenAI Agents SDK, accept the loop, the tools, the memory, the orchestration, the

safetyharrison-chase--x

ANTHROPIC VALUATION: 965B WALMART VALUATION: 940B ANTHROPIC REVENUE: 20B WALMART REVENUE: 725B BUT AI IS NOT A BUBBLE, RIGHT?

ANTHROPIC VALUATION: 965B WALMART VALUATION: 940B ANTHROPIC REVENUE: 20B WALMART REVENUE: 725B BUT AI IS NOT A BUBBLE, RIGHT? Media JUST IN: Anthropic raises 65B at 965B valuation 70% chance of IPO th

AOE: Exhaustive Out-of-Distribution Detection via Recalibrating Outlier Labels

arXiv:2605.28021v1 Announce Type: new Abstract: Out-of-distribution (OOD) detection is essential for deploying machine learning models in open-world and safety-critical scenarios, where test inputs ma

AREA: Attribute Extraction and Aggregation for CLIP-Based Class-Incremental Learning

arXiv:2605.28809v1 Announce Type: new Abstract: Class-Incremental Learning (CIL) is important in building real-world learning systems. In CLIP-based CIL, the model performs classification by comparing

Artemis: Structured Visual Reasoning for Perception Policy Learning

arXiv:2512.01988v2 Announce Type: replace Abstract: Recent reinforcement-learning frameworks for visual perception policy usually incorporate intermediate reasoning chains expressed in natural languag

Auditable Decision Models with Learned Abstention and Real-Time Steering

arXiv:2605.27768v1 Announce Type: new Abstract: Production AI systems often operate with incomplete, conflicting, or insufficient evidence. Forced classifiers collapse such cases into action labels, w

Auditing Stance Asymmetry in Generative Explanations

arXiv:2605.27988v1 Announce Type: new Abstract: Bias evaluation for language models has made substantial progress on bounded comparisons, such as overt derogation, stereotype association, or label-sen

Automated Estimation of Impact Time, Impact Location, and Shuttlecock Speed in Badminton Smashes Using Event Cameras

arXiv:2605.28011v1 Announce Type: new Abstract: Quantifying impact phenomena in badminton smashes is important for evaluating both athletic performance and equipment; however, conventional measurement

AWS CEO Matt Garman: The idea that AI will replace junior developers is “the dumbest thing I have ever heard.”

AWS CEO Matt Garman stated that the notion AI will replace junior developers is fundamentally misguided, suggesting instead that AI tools will augment and enhance developer productivity rather than el

Bayesian Deployment Approval for Learned Landing Controllers under Finite Rollout Validation

arXiv:2605.27720v1 Announce Type: new Abstract: Reinforcement learning and data-driven autonomous controllers are commonly evaluated using cumulative reward and empirical success frequency under finit

Bayesian Gated Non-Negative Contrastive Learning

arXiv:2605.28441v1 Announce Type: cross Abstract: While Contrastive Learning (CL) has revolutionized self-supervised representation learning, its latent representations remain highly entangled and opa

Becoming an AI-native company is existential for every business, says Dell CMO

Becoming an AI-native company is no longer the competitive advantage it was five minutes ago. Now, it’s an empirical obligation, and Dell Technologies Inc. made that case to customers at its annual fl

safetysiliconangle

Behavioural Analysis of Alignment Faking

arXiv:2605.27681v1 Announce Type: new Abstract: Alignment faking (AF) refers to a model strategically complying with a training objective to avoid behavioural modification while preserving its deploym

Beyond Binary: Sim-to-Real Dexterous Manipulation with Physics-Grounded Contact Representation

arXiv:2605.28812v1 Announce Type: cross Abstract: A primary bottleneck in contact-rich manipulation is the difficulty of collecting real-world data. Sim-to-real reinforcement learning offers a scalabl

BiasEdit: A Training-Free Bias-Detect-and-Edit Framework for Learning Fair Visual Classifiers

arXiv:2605.28450v1 Announce Type: cross Abstract: Visual data from the Web power image classifiers, which often underpin many web services, such as recommendation and content moderation. However, the

Blind PRNG Hijacking: An Undetectable Integrity-Preserving Attack Against LLM Watermarking

arXiv:2605.28632v1 Announce Type: cross Abstract: Cryptographic watermarking is a leading defense for attributing text generated by large language models (LLMs). Existing schemes, including KGW, Unigr

Boundary Suppression Asymmetry in Post-trained Assistants: Over-expansion as a Controllability Cost

arXiv:2605.27969v1 Announce Type: new Abstract: Post-trained language-model assistants are often optimized to avoid under-answering, encouraging complete, helpful, cautious, and proactive responses. W

BPPO: Binary Prefix Policy Optimization for Efficient GRPO-Style Reasoning RL with Concise Responses

arXiv:2605.28028v1 Announce Type: new Abstract: Group Relative Policy Optimization (GRPO) is widely used for training reasoning models, but updating all sampled completions in each group incurs substa

Breaking the Script Barrier: Enabling Automatic Alignment for PoS-based ASR Error Analysis in Non-Latin Scripts

arXiv:2605.28438v1 Announce Type: new Abstract: Automatic Speech Recognition (ASR) systems are commonly evaluated using aggregate metrics such as Word Error Rate (WER), which do not capture the lingui

Bridging the Detection-to-Abstention Gap in Reasoning Models under Insufficient Information

arXiv:2605.28070v1 Announce Type: new Abstract: We highlight a failure mode of large reasoning models on questions with insufficient information: models may recognize that a problem is under-specified

Calibrated Inference for the Conditional Average Treatment Effect in the Few-Placebo Regime via Gaussian Processes

arXiv:2605.27473v1 Announce Type: cross Abstract: Estimating how much an intervention helps a given individual the conditional average treatment effect (CATE) is increasingly central to decision-makin

CANDOR: Counterfactual ANnotated DOubly Robust Off-Policy Evaluation

arXiv:2412.08052v2 Announce Type: replace Abstract: Off-policy evaluation (OPE) is critical for applying contextual bandit algorithms to high-stakes decision-making settings such as healthcare, where

Causal Direct Preference Optimization for Distributionally Robust Generative Recommendation

arXiv:2603.22335v2 Announce Type: replace-cross Abstract: Direct Preference Optimization (DPO) guides large language models (LLMs) to generate recommendations aligned with user historical behavior dis

Causal Machine Learning: A Survey and Open Problems

arXiv:2206.15475v3 Announce Type: replace Abstract: Causal Machine Learning (CausalML) is an umbrella term for machine learning methods that formalize the data-generation process as a structural causa

Chance-Constrained MPPI under State and Dynamic Object Prediction Uncertainty and the Evaluation of Collision Risk Calibration

arXiv:2605.28330v1 Announce Type: new Abstract: Chance-constrained Model Predictive Path Integral (MPPI) control is increasingly adopted for navigation in dynamic environments to explicitly bound coll

safetyarxiv-cs-ro

CIRF: Tokenizing Chain-of-Thoughts into Reusable Functional Units for Efficient Latent Reasoning in Large Language Models

arXiv:2605.28292v1 Announce Type: new Abstract: Implicit Chain-of-Thought (CoT) reduces the inference cost of large language models by internalizing the explicit rationales. However, existing approach

CodeGENCAT: Generative Computerized Adaptive Testing for Open-ended Coding Problems

arXiv:2602.20020v2 Announce Type: replace Abstract: Existing Computerized Adaptive Testing (CAT) frameworks typically select questions based on the predicted likelihood that the student will answer co

Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems

arXiv:2602.15198v2 Announce Type: replace-cross Abstract: Multi-agent systems, where LLM agents communicate through free-form language, enable sophisticated coordination for solving complex cooperativ

Commit to the Bit: Reactive Reinforcement Learning Done Right

arXiv:2605.28276v1 Announce Type: new Abstract: Reinforcement learning algorithms are commonly analyzed (and designed) under the Markov assumption. This is unrealistic, as most environments encountere

Compositional Text-to-Image Generation Via Region-aware Bimodal Direct Preference Optimization

arXiv:2605.28615v1 Announce Type: new Abstract: Despite the rapid progress of text-to-image (T2I) models, generating images that accurately reflect complex compositional prompts (covering attribute bi

Con-DSO: Learning Short-Horizon Consistency Priors for RGB-D Direct Sparse Odometry

arXiv:2605.27952v1 Announce Type: new Abstract: Visual odometry (VO) is a fundamental component in robotics and augmented reality. RGB-D direct VO benefits from metric depth measurements, but it can d

COTTA: Context-Aware Transfer Adaptation for Trajectory Prediction in Autonomous Driving

arXiv:2604.00402v2 Announce Type: replace-cross Abstract: Developing robust models to accurately predict the trajectories of surrounding agents is fundamental to autonomous driving safety. However, mo

Counterfactually Fair Regression via Optimal Transport

arXiv:2605.28251v1 Announce Type: cross Abstract: We consider the problem of learning a counterfactually fair regressor. We adopt a causal uncertainty view in which counterfactual fairness is defined

CPPO: Contrastive Perception Policy Optimization for VLM Agents

arXiv:2601.00501v2 Announce Type: replace Abstract: We introduce CPPO, a Contrastive Perception Policy Optimization method for finetuning vision--language models (VLMs). Reliable perception is a core

CRaFT: Circuit-Guided Refusal Feature Selection via Cross-Layer Transcoders

arXiv:2604.01604v2 Announce Type: replace Abstract: While modern LLMs are aligned to refuse harmful requests, it is essential to understand the underlying mechanistic basis of this refusal behavior fo

crazy that this was announced on the day tokenmaxxing died.

crazy that this was announced on the day tokenmaxxing died. Anthropic raised 65 billion in a funding round that valued the artificial intelligence company at 965 billion including the new investment,

Cross-Entropy Games and Frost Training

arXiv:2605.27701v1 Announce Type: new Abstract: We present Frost Training, a method for improving Monte Carlo-based policy optimization for a large family of LLM-as-a-judge tasks called Cross-Entropy