AI news archive.
19,856 articles, newest first, from the outlets listed on the sources page.
- arXiv cs.AIresearchVLA Policies Do Not Know When Not to Act: Unintended Task Execution and Start Suppression
arXiv:2610.41609v1 Announce Type: cross Abstract: A robot should act only on an instruction from a person or on an intent that the robot explicitly represents and generates. Unintended actions taken without such an instr…
- arXiv cs.AIresearchAttention Maps Know When to Think: Training-Free Difficulty-Aware Chain-of-Thought for Reasoning Segmentation
arXiv:2610.41618v1 Announce Type: cross Abstract: Chain-of-Thought (CoT) prompting has become a standard reasoning strategy in Multi-modal Large Language Models (MLLMs), showing strong performance in complex vision-langu…
- arXiv cs.AIresearchGRIT: Gradient Masking for Visually Misgrounded Tokens in Text-Rich Reasoning
arXiv:2610.41629v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has recently been extended to vision-language models (VLMs), yet its effectiveness remains limited for text-rich rea…
- arXiv cs.AIresearchWhen Steering Shifts Discourse: Identifying and Intervening on Discourse-Level Changes due to Activation Steering in LLMs
arXiv:2610.41661v1 Announce Type: cross Abstract: Activation steering is an inference-time technique used to control or modify the behavior, tone, or safety properties of a large language model (LLM) without retraining t…
- arXiv cs.AIresearchOT-CAL: Optimal-Transport-Guided Adaptive Sample Selection for Efficient Visual Learning
arXiv:2610.41677v1 Announce Type: cross Abstract: Deep neural networks for image recognition are typically trained using mini-batches, with a forward pass to compute predictions and losses followed by a more computationa…
- arXiv cs.AIresearchWhen the Past Shapes Action: Understanding Embodied Memory
arXiv:2610.41688v1 Announce Type: cross Abstract: Recent advances in generative foundation models have expanded the capabilities of embodied AI, enabling increasingly general navigation, manipulation, and long-horizon ta…
- arXiv cs.AIresearchSimple Features Beat Most AI Control Monitors
arXiv:2610.41713v1 Announce Type: cross Abstract: AI control relies on monitors to catch a capable model that sabotages its work. These monitors are evaluated on their ability to differentiate honest from backdoored code…
- arXiv cs.AIagentsLatent-Lens: Visual Perception in Small Language Models Through Latent Communication
arXiv:2610.41751v1 Announce Type: cross Abstract: Bridging independent vision-language models (VLMs) and small language models (SLMs) in a modular multi-agent system typically forces them to exchange information as discr…
- arXiv cs.AIresearchDifferent Models, Shared Failures: Auditing Frozen-Encoder Pools for Medical Segmentation
arXiv:2610.41790v1 Announce Type: cross Abstract: Average segmentation accuracy and model count do not reveal whether models fail on the same cases. We present CoFailMed, an audit of correlated failures in pretrained-enc…
- arXiv cs.AIresearchImplicit Assumptions About Tasks in Continual Learning
arXiv:2610.41792v1 Announce Type: cross Abstract: Continual learning is the study of how a model accumulates knowledge from data received over time, rather than all at once. In practice, this is typically achieved by div…
- arXiv cs.AIresearchAccurate and Auditable ECG Classification via Visibility Graph Neural Networks
arXiv:2610.41795v1 Announce Type: cross Abstract: Physicians read electrocardiograms (ECGs) from waveform shape and timing, yet accurate deep-learning classifiers rarely show which waveform regions drive their prediction…
- arXiv cs.AIresearchBackpropagation-free Deep Learning for Time-Series Classification
arXiv:2610.41805v1 Announce Type: cross Abstract: Deep neural networks for time series classification (TSC) are trained almost exclusively by error backpropagation, which imposes update locking and the weight-transport p…
- arXiv cs.AIagentsSimXAI: A Simple Drop-in Layer Increases Interaction Quality and Understanding in Conversational XAI
arXiv:2610.41856v1 Announce Type: cross Abstract: Conversational explainable AI (ConvXAI) agents often use GUS-style architectures with separate modules for intent recognition, dialogue management, and template-based res…
- arXiv cs.AIresearchAI evaluators favor job applications written by certain AI models, independent of candidate quality
arXiv:2610.41866v1 Announce Type: cross Abstract: Job applicants increasingly use large language models (LLMs) to write cover letters, while employers deploy LLMs to screen them. In this all-AI setting, we ask whether th…
- arXiv cs.AIresearchGeneral-purpose foundation models surpass specialized ones on chemical property prediction tasks
arXiv:2610.41899v1 Announce Type: cross Abstract: Foundation models (FMs) learn reusable representations from large collections of unlabeled data. In chemistry the corresponding data mainly consists of molecular structur…
- arXiv cs.AIresearchObservable-Specific Validity Tests for Scientific AI: Controlled Approximations and Intervention Response
arXiv:2610.41939v1 Announce Type: cross Abstract: An approximation can reproduce an energy while missing its response, or satisfy an error tolerance without being certified by a given bound. We formulate observable-speci…
- arXiv cs.AIresearchAligning MLLMs with the Latent Structure of Human Cognition via Behavior-Derived Semantic Dimensions
arXiv:2610.41951v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) have achieved impressive performance across a wide range of reasoning tasks, yet it remains unclear whether their internal repres…
- arXiv cs.AIresearchChronos-2: A foundation model for multivariate time series forecasting
arXiv:2610.41956v1 Announce Type: cross Abstract: Accurate forecasting with uncertainty quantification is critical in a variety of scientific applications ranging from epidemiology and economics to energy systems and ear…
- arXiv cs.AIagentsSecPO: Principled Adversarial Training for Prompt Injection Security
arXiv:2610.41998v1 Announce Type: cross Abstract: Recognized as a top threat to AI agents, prompt injection allows instructions from untrusted data (email, documents, websites) to manipulate the agent's operation. The ex…
- arXiv cs.AIresearchAutoAnalysis: An automated, self-updating meta-analysis of human-AI collaboration
arXiv:2610.42001v1 Announce Type: cross Abstract: As generative artificial intelligence (AI) becomes more capable and widely used, understanding when and how humans should work with these systems is becoming increasingly…
- arXiv cs.AIresearchKnow When the Equation Is Wrong: Structural Uncertainty for LLM-Guided Symbolic Regression under Extrapolation
arXiv:2610.42024v1 Announce Type: cross Abstract: Large language models (LLMs) can help discover mathematical equations from data, but most LLM-guided symbolic regression methods focus on a single best equation without q…
- arXiv cs.AIresearchThere Is No Free Loop: When Does Looping a Trained Model Pay for Its Inference Compute?
arXiv:2610.42030v1 Announce Type: cross Abstract: Looping re-applies a block of layers so that each token passes through the block more than once. Adding a loop to a trained model is attractive because it leaves the para…
- arXiv cs.AIresearchLearning Clean Generative Processes from Noisy Measurements
arXiv:2610.42036v1 Announce Type: cross Abstract: We study how to learn a generative model of clean images using only noisy observations. Diffusion processes and their generalizations to non-Gaussian noise require poster…
- arXiv cs.AIresearchNear-Optimal Second-Order Convex Finite-Sum Optimization under Individual Smoothness
arXiv:2610.42040v1 Announce Type: cross Abstract: We study the stochastic second-order oracle complexity of solving smooth convex finite-sum optimization with $n$ components. Very recently, Tsedenov \& Kovalev (2026) sho…
- arXiv cs.AIresearchCRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness
arXiv:2610.42054v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has become the standard method for training language models to reason. However, RLVR typically relies solely on outc…
- arXiv cs.AIresearchSubquadratic Attention with a Hierarchical Tree-Structured Indexer
arXiv:2610.42112v1 Announce Type: cross Abstract: The quadratic scaling of attention is a fundamental bottleneck limiting the long-context capabilities of large language models (LLMs). A variety of approaches have attemp…
- arXiv cs.AIagentsRoboArchitect: Automating the Design of Robotic Agents with Persistent Memory
arXiv:2610.42141v1 Announce Type: cross Abstract: Recent work shows that a frozen robot policy performs far better inside an agent harness, where control code decides when to query the policy, consult vision tools, or co…
- arXiv cs.AIagentsHierarchical World Models with Implicit Dynamics
arXiv:2610.42146v1 Announce Type: cross Abstract: Planning with world models enables test-time adaptation for embodied agents, offering superior generalization over direct policy learning. However, long-horizon tasks rem…
- arXiv cs.AIresearchShared Modular Recurrence in Contextual MDPs for Universal Morphology Control
arXiv:2506.08630v4 Announce Type: replace Abstract: A universal controller for any robot morphology would greatly improve computational and data efficiency. Steps have been made towards such multi-robot control by utiliz…
- arXiv cs.AIresearchA Hybrid SMT-NRA Solver: Integrating 2D Cell-Jump-Based Local Search, MCSAT and OpenCAD
arXiv:2507.00557v3 Announce Type: replace Abstract: In this paper, we propose a hybrid framework for Satisfiability Modulo the Theory of Nonlinear Real Arithmetic (SMT-NRA for short). First, we introduce a two-dimensiona…
- arXiv cs.AIagentsAx-Prover: A Deep Reasoning Agentic Framework for Theorem Proving in Mathematics and Quantum Physics
arXiv:2510.12787v5 Announce Type: replace Abstract: We present Ax-Prover, a multi-agent system for automated theorem proving in Lean that can solve problems across diverse scientific domains and operate either autonomous…
- arXiv cs.AIresearchThinkSafe: Self-Generated Safety Alignment for Reasoning Models
arXiv:2601.23143v5 Announce Type: replace Abstract: Large reasoning models achieve strong performance through reinforcement learning but can suffer safety degradation. External teacher distillation can restore safety yet…
- arXiv cs.AIagentsSoK: Agentic Retrieval-Augmented Generation (RAG): Taxonomy, Architectures, Evaluation, and Research Directions
arXiv:2603.07379v2 Announce Type: replace Abstract: Retrieval-Augmented Generation (RAG) systems are increasingly evolving into agentic architectures where large language models autonomously coordinate multi-step reasoni…
- arXiv cs.AIagentsControllable User Simulation
arXiv:2605.11519v2 Announce Type: replace Abstract: Using offline datasets to evaluate conversational agents often fails to cover rare scenarios or to support testing new policies. This has motivated the use of controlla…
- arXiv cs.AIresearchTeaching Large Language Models When Not to Know: Learning Temporal Critique for Ex-Ante Reasoning
arXiv:2605.14636v2 Announce Type: replace Abstract: Large language models (LLMs) often fail to reason under temporal cutoffs: when prompted to answer from the standpoint of an earlier time, they exploit knowledge that be…
- arXiv cs.AIresearchGENSTRAT: Toward a Science of Strategic Reasoning in Large Language Models
arXiv:2605.23238v2 Announce Type: replace Abstract: Fixed game suites for evaluating strategic reasoning in large language models (LLMs) saturate and are hard to keep free of contamination. We introduce GENSTRAT, a syste…
- arXiv cs.AIagentsPhysics-Audited Agentic Discovery in Scientific Machine Learning
arXiv:2607.07379v2 Announce Type: replace Abstract: Agentic discovery can yield mechanics surrogates with low validation error that nonetheless violate physical requirements. We present Physics-Audited Agentic SciML (PA-…
- arXiv cs.AIresearchFedPA-LoRA: Product-Aligned Framework for Mitigating Aggregation and Initialization Errors in Heterogeneous Federated LoRA
arXiv:2608.15381v2 Announce Type: replace Abstract: Low-Rank Adaptation (LoRA) enables efficient federated fine-tuning of large language models, but its factorization creates a tension between aggregation fidelity and lo…
- arXiv cs.AIresearchFrom Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention Mechanism
arXiv:2608.21702v2 Announce Type: replace Abstract: Retrieval-Augmented Generation (RAG) grounds LLM generation on retrieved documents, but the standard terminal retrieval stage--dense-vector similarity, optionally follo…
- arXiv cs.AIresearchSIRF: A Spec-Internalized Risk Foundation Model for Industrial Content Risk Control
arXiv:2609.11752v2 Announce Type: replace Abstract: For industrial content risk control, the real deployment constraint is not average accuracy but how much risk can be auto-handled under high precision and second-level…
- arXiv cs.AIresearchA Unified Per-Token Gating Family for On-Policy Distillation: FKL/RKL Mixing with Multi-Channel and Bias Coefficients
arXiv:2609.11768v2 Announce Type: replace Abstract: Per-token gating of forward/reverse KL losses has become a standard technique for on-policy knowledge distillation (OPD), but existing methods such as EOPD (Jin et al.,…
- arXiv cs.AIresearchTime Will Tell: Harness Time to Denoise Collaborative Filtering
arXiv:2411.09181v2 Announce Type: replace-cross Abstract: Neural graph collaborative filtering is a powerful tool for recommendation due to its superb capability of capturing the user-item interaction. Despite the substa…
- arXiv cs.AIresearchIntegrating AI for Human-Centric Breast Cancer Diagnostics: A Multi-Scale and Multi-View Swin Transformer Framework
arXiv:2503.13309v3 Announce Type: replace-cross Abstract: Despite advancements in Computer-Aided Diagnosis (CAD) systems, breast cancer remains one of the leading causes of cancer-related deaths among women worldwide. Re…
- arXiv cs.AIresearchBeyond Public Access in LLM Pre-Training Data
arXiv:2505.00020v3 Announce Type: replace-cross Abstract: Using a legally obtained dataset of 34 copyrighted O'Reilly Media books (13,962 paragraphs), we apply the DE-COP membership inference attack to test whether OpenA…
- arXiv cs.AIresearchOptimizing Anytime Reasoning via Budget Relative Policy Optimization
arXiv:2505.13438v4 Announce Type: replace-cross Abstract: Scaling test-time compute is crucial for enhancing the reasoning capabilities of large language models (LLMs). Existing approaches typically employ reinforcement…
- arXiv cs.AIresearchLLINBO: Trustworthy LLM-in-the-Loop Bayesian Optimization
arXiv:2505.14756v3 Announce Type: replace-cross Abstract: Bayesian optimization (BO) is a sequential decision-making tool widely used for optimizing expensive black-box functions. Recently, Large Language Models (LLMs) h…
- arXiv cs.AIresearchFully Offline Reinforcement Learning
arXiv:2505.22442v4 Announce Type: replace-cross Abstract: Offline RL (ORL) promises safe and sample-efficient deployment but existing methods rely on undocumented online interactions for hyperparameter tuning and lack re…
- arXiv cs.AIresearchVisualSphinx: Large-Scale Synthetic Vision Logic Puzzles for RL
arXiv:2505.23977v2 Announce Type: replace-cross Abstract: Vision language models (VLMs) are expected to perform effective multimodal reasoning and make logically coherent decisions, which is critical to tasks such as dia…
- arXiv cs.AIresearchPPS: Plug-and-Play Saccadic Vision for Fine-Grained Classification
arXiv:2509.15688v2 Announce Type: replace-cross Abstract: Fine-grained visual classification benefits from high-resolution inputs to resolve subtle, localized cues, but at the cost of large compute and memory budgets. In…
- arXiv cs.AIresearchMLLMEraser: Achieving Test-Time Unlearning in Multimodal Large Language Models through Activation Steering
arXiv:2510.04217v4 Announce Type: replace-cross Abstract: Multimodal large language models (MLLMs) have demonstrated remarkable capabilities across vision-language tasks, yet their large-scale deployment raises pressing…