AI news archive.
19,864 articles, newest first, from the outlets listed on the sources page.
- arXiv cs.AIresearchA Gravitational Interpretation of Safety Reversion under Fine-Tuning
arXiv:2606.28525v2 Announce Type: replace-cross Abstract: Safety alignment in large language models can degrade during post-training even when neither the data nor the objective is intentionally adversarial. Alignment re…
- arXiv cs.AIresearchReal-Time Hard Negative Sampling via LLM-based Clustering for Large-Scale Two-Tower Retrieval
arXiv:2607.00448v3 Announce Type: replace-cross Abstract: The two-tower model is widely used in the retrieval stage of large-scale recommendation systems, where training typically relies on in-batch and/or out-of-batch n…
- arXiv cs.AIresearchFlow-Map GRPO: Reinforcement Learning for Few-Step Flow-Map Generators via Anchored Stochastic Composition
arXiv:2607.00535v2 Announce Type: replace-cross Abstract: Few-step flow-map generators, such as consistency models and MeanFlow, accelerate sampling by learning long-range transport maps between noise and data. However,…
- arXiv cs.AIresearchScaling Laws for Collapse in Asynchronous GRPO
arXiv:2607.01083v3 Announce Type: replace-cross Abstract: Asynchronous reinforcement learning improves the throughput of large language model post-training by decoupling rollout generation from policy optimization, but i…
- arXiv cs.AIresearchSampling Meets Interaction: Sequentially-Controlled Multi-Particle Flow-Maps for Efficient Inference-time Search
arXiv:2607.01144v2 Announce Type: replace-cross Abstract: While generative models have enabled training-free reward alignment, existing particle-based methods are fundamentally constrained by their propensity to local ex…
- arXiv cs.AIresearchLACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning
arXiv:2607.02513v2 Announce Type: replace-cross Abstract: LLMs memorize sensitive training data, including personally identifiable information (PII), creating a pressing need for reliable post hoc removal methods. Unlear…
- arXiv cs.AIresearchBFMT: Enhancing Search Capabilities of Tree Sampler via Bootstrap Flow-Map Tree
arXiv:2607.02915v2 Announce Type: replace-cross Abstract: The ability to efficiently navigate high-dimensional spaces to identify optimal candidates-whether designing targeted molecular structures or generating images wi…
- arXiv cs.AIresearchOne Image, No Tokens: A Controlled Study of Glyph-Based Chinese Language Modeling
arXiv:2607.03994v2 Announce Type: replace-cross Abstract: Modern language models generally represent text as sequences of discrete token embeddings, an assumption deeply rooted in current practice but rarely questioned.…
- arXiv cs.AIagentsJet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE
arXiv:2607.07740v4 Announce Type: replace-cross Abstract: Modern LLMs are increasingly deployed in long-context applications such as retrieval-augmented generation, repository-level coding, and agentic workflows whose ac…
- arXiv cs.AIresearchSource-Lifted Flow Matching for Intervenable Multimodal Imitation
arXiv:2607.10206v2 Announce Type: replace-cross Abstract: Flow-matching policies are promising for imitation learning because they model complex multimodal action distributions. However, their stochasticity is largely pa…
- arXiv cs.AIagentsTowards Autonomous and Auditable Medical Imaging Model Development
arXiv:2607.10522v2 Announce Type: replace-cross Abstract: Large language model (LLM) agents are beginning to automate machine learning engineering (MLE) by coupling planning, code execution, debugging, and empirical feed…
- arXiv cs.AIagentsAgent Hacks Agents: Autoresearch Discovers Vulnerabilities in Production Agents
arXiv:2607.11698v2 Announce Type: replace-cross Abstract: Production LLM agents such as Claude Code and Codex can modify files and execute commands, so safety failures become real destructive actions. Automatic red-teami…
- arXiv cs.AIresearchThe Entanglement Wall: Activation-Space Probes as Risk Detectors, Not Context Adjudicators
arXiv:2607.13075v2 Announce Type: replace-cross Abstract: An activation-based safety gate must distinguish harmful assistance from legitimate work on the same topic. We test whether probes trained on broad harmful-versus…
- arXiv cs.AIagentsEta Given Delta: Defining LLM Tool Efficiency With Marginal Tool Utility
arXiv:2607.14108v2 Announce Type: replace-cross Abstract: This paper introduces tool efficiency, a new quantitative metric to evaluate the rate of useful tool calls in an LLM agent trajectory. To ensure that tool efficie…
- arXiv cs.AIresearchWhat Does Model Growth Really Add? Functional Capacity Beyond Parameter Count
arXiv:2607.14571v2 Announce Type: replace-cross Abstract: Model growth adds parameters to a trained checkpoint and continues training, aiming to increase capacity while reusing the existing model instead of training from…
- arXiv cs.AIresearchTRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment
arXiv:2607.16242v2 Announce Type: replace-cross Abstract: Fine-Tuning-as-a-Service (FTaaS) platforms let users perform supervised fine-tuning (SFT) on customized data, but this pipeline can erode model safety alignment.…
- arXiv cs.AIresearchCoCurve: Cross-Module Co-Pruning Curvature for Structured LLM Pruning
arXiv:2607.17568v2 Announce Type: replace-cross Abstract: Resource-constrained deployment requires sustaining large language model (LLM) capabilities as models scale under fixed memory and computation budgets. Structured…
- arXiv cs.AIresearchCriPO: Enhancing Rubric-based RL via Self-Distillation
arXiv:2607.18082v4 Announce Type: replace-cross Abstract: Rubric-based Reinforcement Learning (RL) has recently shown promise in improving Large Language Models (LLMs) on open-ended tasks. A widely recognized limitation…
- arXiv cs.AIresearchMOPDA: Mixed-Trajectory On-Policy Distillation for Language-Guided Industrial Anomaly Detection
arXiv:2607.18850v2 Announce Type: replace-cross Abstract: Large vision-language models (LVLMs) have shown strong potential for industrial anomaly detection (IAD) by providing image-level anomaly judgments and interpretab…
- arXiv cs.AIresearchSimulating Eutopia: Revisiting Long-term Fairness with Outcomes, Performativity, and Dynamics
arXiv:2607.19389v2 Announce Type: replace-cross Abstract: As AI-driven Decision Makers (ADMs) influence our socioeconomic reality, their roles in both enhancing efficiency and amplifying the social biases have drawn sign…
- arXiv cs.AIresearchConfidently Deceptive: On the Relationship Between Confidence and Deception in LLMs
arXiv:2607.20444v2 Announce Type: replace-cross Abstract: The increasing capabilities of large language models (LLMs) are being accompanied by deep-rooted risks of deceptive behaviours that cause models to produce mislea…
- arXiv cs.AIresearchInteraction Dynamics Modeling and Predictive Control for Safe Steerable Catheter--Tissue Interaction
arXiv:2607.20939v3 Announce Type: replace-cross Abstract: Steerable catheters are the primary tool for cardiac electrophysiology (EP) procedures including radiofrequency ablation, where the tip must be positioned precise…
- arXiv cs.AIresearchWhy Large Language Models and Humans Converge and Diverge in Evaluating Creativity
arXiv:2607.22218v2 Announce Type: replace-cross Abstract: Despite the growing use of large language models (LLMs) as creativity evaluators, evidence of their alignment with human evaluations remains mixed, raising the qu…
- arXiv cs.AIresearchDualityCert: Verifier-Gated Language-Model Repair of Broken Duality Claims in Quantum Field Theory
arXiv:2607.23614v3 Announce Type: replace-cross Abstract: We present DualityCert, a symbolic verifier for candidate Seiberg-duality claims in four-dimensional N=1 quiver gauge theories. The verifier evaluates 't Hooft an…
- arXiv cs.AIagentsOne Analyst Is Not Ground Truth: Grading Agent-Built Financial Models Against Observed Professional Practice
arXiv:2607.24889v2 Announce Type: replace-cross Abstract: Language model agents can now construct complete financial models, but it remains unclear whether they have learned the financial judgment that gives those models…
- arXiv cs.AIresearchPost-Training at the Edge of Detectability: A Game-Theoretic Approach to Fine-Tuning
arXiv:2607.26358v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) fine-tuning is widely used in language model training to improve performance on a target task while limiting drift from a reference po…
- arXiv cs.AIresearchBudget-Aware LLM Discovery via Cost-Calibrated Frontier Utility
arXiv:2607.26828v4 Announce Type: replace-cross Abstract: Large language models increasingly support scientific and algorithmic discovery through inference-time search over evaluated candidates. Existing adaptive discove…
- arXiv cs.AIresearchSciFigQual-Bench: A Benchmark for Scientific Figure Quality Assessment with Full-Manuscript Context
arXiv:2607.27084v3 Announce Type: replace-cross Abstract: Scientific figure quality is bound to the manuscript: a crop can be visually clear and still contradict its caption or the paragraph that cites it. Most image qua…
- arXiv cs.AIresearchRedFlow: Redirect Failure into Action-level Corrections for Flow-matching VLA Policy
arXiv:2607.27782v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) can improve Vision-Language-Action (VLA) policies from deployment experience, but reward- and preference-based RL primarily identifies…
- arXiv cs.AIresearchEfficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates
arXiv:2607.28959v2 Announce Type: replace-cross Abstract: Adversarial training is one of the most effective defenses against adversarial attacks, yet the computational cost remains prohibitive at modern scales, especiall…
- arXiv cs.AIagentsDynActiveGS: Active Gaussian Splatting for Dynamic Scene Reconstruction
arXiv:2608.01178v2 Announce Type: replace-cross Abstract: We present DynActiveGS, a dynamic-aware active reconstruction framework based on 3D Gaussian Splatting (3DGS) for autonomous exploration in dynamic environments.…
- arXiv cs.AIresearchPreferred, Not Safer: Pairwise Preference Is a Poor Proxy for Clinical Safety
arXiv:2608.02617v2 Announce Type: replace-cross Abstract: We evaluate whether clinician pairwise preferences provide a reliable signal of clinical safety in large language model (LLM) evaluation using expert feedback fro…
- arXiv cs.AIresearchdots.tts.edit: Precisely Controlled Speech Editing with a Continuous Autoregressive Model
arXiv:2608.02673v4 Announce Type: replace-cross Abstract: Speech editing for content creation requires precise control over both what an edit should do and where it should apply. Free-form natural language provides a fle…
- arXiv cs.AIagentsE$^3$-Orch: Towards Effective, Efficient, and Extensible Agentic Orchestration with Reinforcement Learning
arXiv:2608.04588v2 Announce Type: replace-cross Abstract: Agentic orchestration enables multiple autonomous agents to solve complex tasks through adaptive decomposition, delegation, and execution. However, existing orche…
- arXiv cs.AIresearchSSTQ:Privacy-Preserving Vector Quantization via Subsampled Stochastic TurboQuant
arXiv:2608.05127v3 Announce Type: replace-cross Abstract: Achieving local differential privacy in distributed optimization while maintaining low communication cost remains challenging. Existing vector quantization method…
- arXiv cs.AIresearchOnce a Response, Always a Response: Detecting LLM-generated Text via Latent Prompt Restoration
arXiv:2608.05741v2 Announce Type: replace-cross Abstract: Large language models (LLMs) can generate fluent and convincing text at scale, creating growing risks for misinformation dissemination, educational misuse, and pl…
- arXiv cs.AIresearchMultivariate Time Series Forecasting needs Cross Variable Loss
arXiv:2608.05742v4 Announce Type: replace-cross Abstract: Multivariate time series forecasting presents unique challenges because future variables often co-evolve under shared system dynamics. While existing studies main…
- arXiv cs.AIresearchFastKron: Efficient Quantization with Kronecker-Factored Hessians
arXiv:2608.06291v2 Announce Type: replace-cross Abstract: We accelerate a family of algorithms for neural network quantization which utilize a two-sided version of the GPTQ/LDLQ algorithm. Standard GPTQ-style adaptive ro…
- arXiv cs.AIresearchTransSLR: A Lightweight Transformer for Sign Language Recognition
arXiv:2608.06407v2 Announce Type: replace-cross Abstract: Automated sign language recognition for underrepresented languages remains a largely unsolved problem. Central African Sign Language (CASL) exemplifies this chall…
- arXiv cs.AIresearchMaskFlow: Precise, Consistent and Seamless Regional Image Editing
arXiv:2608.06929v3 Announce Type: replace-cross Abstract: Regional image editing has attracted considerable attention for its spatial controllability. Although instruction-based and mask-based editing methods can achieve…
- arXiv cs.AIresearchEvidence-RL: Towards Evidence-intensive Visual Reasoning
arXiv:2608.08021v2 Announce Type: replace-cross Abstract: Vision-Language Models (VLMs) should answer from concrete image evidence rather than language priors, dataset shortcuts, or irrelevant visual context. Existing pe…
- arXiv cs.AIresearchRL-Native Distillation: Exploiting Scored Trajectories for Few-Step Image Generation
arXiv:2608.09226v2 Announce Type: replace-cross Abstract: Efficient text-to-image generation requires both reinforcement-learning (RL)-based reward alignment and few-step distillation, yet these procedures are typically…
- arXiv cs.AIagentsRefineFly: Failure-Aware Post-Training for Aerial Vision-Language Navigation
arXiv:2608.09467v3 Announce Type: replace-cross Abstract: Unmanned aerial vehicle vision-language navigation (UAV-VLN) requires agents to translate visual observations and language instructions into reliable flight actio…
- arXiv cs.AIresearchError-Aware Reverse Auction Mechanism for Large Language Model Routing
arXiv:2608.12719v2 Announce Type: replace-cross Abstract: Routing each query to a cost-effective large language model (LLM) is critical for balancing quality and cost, yet most routers rely on a centralized task center t…
- arXiv cs.AIagentsERSkill: Evolving for Skill-Guided Adaptive Memory Retrieval
arXiv:2608.12720v2 Announce Type: replace-cross Abstract: While Large Language Model (LLM) agents increasingly rely on long-term memory for persistent interactions, the retrieval mechanisms governing this memory are rare…
- arXiv cs.AIagentsAQuA: Recursively Self-Improving Quantitative Trading Research Agents
arXiv:2608.12841v3 Announce Type: replace-cross Abstract: We study recursive self-improvement at the level of quantitative-investment research: whether an autonomous system can use evidence from earlier experiments to im…
- arXiv cs.AIresearchSpectral Saliency for Machine Unlearning
arXiv:2608.15548v2 Announce Type: replace-cross Abstract: Machine unlearning (MU) aims to remove the influence of specific training data while preserving model utility. As the name suggests, MU can be viewed as the inver…
- arXiv cs.AIresearchMultinomial Subset Routing with Sum-Max Rewards and Operational Constraints
arXiv:2608.16375v2 Announce Type: replace-cross Abstract: We study online routing to subsets of experts under aggregate bandit feedback and long-run operational constraints. Expert contributions can be complementary: for…
- arXiv cs.AIresearchPTXBench: Benchmarking and Adapting LLMs for GPU Kernel Optimization with Architecture-specific PTX
arXiv:2608.17379v3 Announce Type: replace-cross Abstract: We introduce PTXBench, a benchmark for evaluating and adapting large language models (LLMs) to use architecture-specific PTX for GPU kernel optimization. PTXBench…
- arXiv cs.AIresearchPartition the Support, Reconstruct the Residual: Training-Free Sparse Attention for Video Generation and World Models
arXiv:2608.18484v2 Announce Type: replace-cross Abstract: Block-sparse attention accelerates video transformers by grouping queries and keys and computing attention only between selected groups. However, it faces two cha…