Real-time AI-curated news from 98,704+ articles across 50+ sources. Sentiment analysis, importance scoring, and key takeaways — updated every 15 minutes.
Octopus Protocol automates hardware discovery and control for AI agents through a single command, eliminating the need for manual driver and SDK development. The system uses a five-stage pipeline to detect connected devices, generate typed tools via Model Context Protocol, and deploy live endpoints, reducing hardware onboarding from weeks to 10-15 minutes.
Shepherd is a new runtime substrate that enables meta-agents to supervise and optimize other agents through formalized execution traces, achieving 5x faster forking than Docker and demonstrating measurable improvements in coding assistance, optimization, and reinforcement learning tasks. The open-source system mechanizes core operations in Lean and enables replay, branching, and counterfactual exploration of agent behaviors.
Researchers introduce EXPO, an improved reinforcement learning algorithm for LLM mathematical reasoning that dynamically adjusts KL penalty coefficients and prioritizes moderately difficult problems during training. The method demonstrates significant performance improvements over existing GRPO approaches, achieving a 13.34-point absolute gain on AIME 2025 benchmarks.
Researchers introduce Hypothesis-Driven Deep Research (HDRI), a new AI methodology that uses hypotheses as structural organizing tools rather than mere end products, enabling automated knowledge discovery across domains. The INFOMINER system implementing this framework demonstrates significant improvements in fact density (22.4%), verification confidence (0.92), and research completeness, validated through five case studies achieving 4.46/5.0 quality ratings.
Researchers identify critical honesty failures in Large Language Model unlearning methods, where models hallucinate or behave inconsistently after attempting to forget harmful training data. They propose ReVa, a representation-alignment procedure that significantly improves model honesty by better acknowledging forgotten knowledge while maintaining utility on retained information.
Researchers unveiled KnotBench, a comprehensive benchmark testing vision-language models' ability to reason about knot diagrams, revealing that current models like Claude Opus and GPT-5 struggle fundamentally with spatial reasoning and symbolic operations despite perceiving visual details. The benchmark demonstrates a critical gap between perception and reasoning capabilities, with most tasks scoring near or below random chance, suggesting VLMs lack mechanisms to simulate geometric transformations.
Researchers introduce EnactToM, a benchmark testing whether AI agents can understand and act on others' beliefs in multi-agent embodied environments. Current frontier models achieve 0% on functional theory of mind tasks, revealing a critical gap in AI reasoning capabilities despite performing better on direct belief questions.
Researchers introduce Ambig-DS, a benchmark suite that evaluates how AI data-science agents handle ambiguous task specifications. The benchmark reveals that current agents silently commit to incorrect interpretations rather than flagging underspecified requirements, a critical failure mode masked by clean-looking outputs that fail to achieve intended objectives.
Researchers introduce IndustryBench, a 2,049-item benchmark testing large language models on industrial procurement tasks grounded in Chinese national standards. The study reveals that current LLMs perform poorly on safety-critical industrial applications, with the best models scoring only 2.08/3.0, and that extended reasoning paradoxically increases safety violations by introducing unsupported details into answers.
Researchers introduce MolWorld, a novel AI framework that optimizes molecular structures for drug discovery by modeling actionable pathways between molecules. Unlike existing methods, MolWorld ensures discovered candidates are chemically reachable from known compounds through valid intermediate steps, making them practically viable for lead optimization.
Researchers introduce a learnable approach to commitment depth—the number of primitive actions executed before replanning—in vision-language models for long-horizon reasoning. Their adaptive policy outperforms fixed-depth baselines and surpasses GPT-4.5 and Claude Sonnet on puzzle-solving tasks, achieving higher solve rates with fewer actions.
Researchers have discovered ShadowMerge, a novel poisoning attack that exploits vulnerabilities in graph-based agent memory systems used by LLM agents. The attack achieves a 93.8% success rate by injecting malicious relations that conflict with benign data, enabling attackers to manipulate agent behavior while evading existing security defenses.
Researchers present PROBE, a framework that improves how AI software engineering agents recover from failures by converting runtime telemetry into structured diagnoses and bounded recovery guidance. The system achieves 65% diagnosis accuracy and 21.8% recovery rates on previously unresolved cases, with a prototype deployed at Microsoft showing practical viability without disrupting existing workflows.
Researchers have developed EpiGraph, a comprehensive knowledge graph containing 24,324 entities and 32,009 evidence-grounded triplets from 48,166 peer-reviewed papers to improve AI-driven epilepsy diagnosis and treatment. The accompanying EpiBench benchmark demonstrates that integrating structured clinical knowledge into large language models significantly enhances clinical reasoning, with improvements up to 41% in pharmacogenomic applications.
Researchers apply game-theoretic free energy principles to analyze attention head interactions in large language models, discovering that heads exhibit higher-order redundancy. Their framework enables principled pruning of low-contribution heads, achieving 18% FLOP reduction and 22% throughput improvement in GPT2 with minimal performance degradation.
A new research position argues that enterprises should stop treating large language models as monolithic solutions for all tasks and instead use them primarily for structured data extraction within modular architectures. The paper contends that LLMs have inherent capacity limits for enterprise knowledge needs and proposes delegating computation and storage to specialized components like knowledge bases and symbolic systems for better reliability and cost efficiency.
Researchers introduce Chaintrix, an LLM-augmented smart-contract auditing framework that combines AI analysis with deterministic structural verification to reduce false positives. The system achieves 71.7% recall on high-severity vulnerabilities, outperforming existing AI and static analysis tools by 26 percentage points on OpenAI's EVMbench benchmark.
Researchers introduce NIAgent, a multi-agent AI system that automates end-to-end neuroimaging analysis by enabling specialist agents to collaboratively build and optimize executable programs. The system outperforms conventional static workflows like fMRIPrep by adapting dynamically to data and incorporating hierarchical quality control, addressing a critical bottleneck in clinical biomarker development.
Researchers have identified a compact causal mechanism explaining how large language models can be persuaded to abandon factual knowledge through the manipulation of mid-layer attention heads. The vulnerability operates as a discrete latent switch rather than confidence reduction, with persuasion working by redirecting attention via a rank-one feature built from persuasive keywords, revealing persuasion as a narrow and potentially monitorable circuit.
Researchers introduce Agent-ValueBench, the first comprehensive benchmark designed to measure and evaluate the values embedded in autonomous AI agents rather than just their underlying language models. The study reveals that agent values diverge significantly from LLM values and are shaped more decisively by system harnesses and embedded skills than by traditional model alignment or prompt engineering approaches.
Researchers have created a benchmark to test whether machine learning interatomic potentials can generalize to unseen molecules by learning underlying chemical principles. The study reveals that state-of-the-art models, including foundation models trained on millions of molecules, fail significantly on out-of-distribution examples, with errors often 10x higher than on training data.
Researchers introduce NEXUS, a framework enabling embodied AI agents to learn symbolic constraints for safer decision-making in physical environments. The system addresses the gap between probabilistic language models and the deterministic safety requirements of robotics by decoupling physical feasibility from safety specifications, achieving improved task success while refusing unsafe instructions.
Researchers propose workspace optimization, a novel training approach for AI agents that evolves external structured environments rather than model weights. The DreamTeam multi-agent system demonstrates this concept on ARC-AGI-3 benchmarks, achieving 38.4% accuracy—a 2.4-point improvement over previous state-of-the-art while reducing computational actions by 31%.
Researchers introduced MDGYM, a benchmark testing AI agents' ability to autonomously execute molecular dynamics simulations, finding that even the strongest systems solve only 21% of easy tasks. The poor performance reveals that advanced code generation does not translate to physical reasoning, exposing a critical gap between general software engineering competence and domain-specific scientific workflows.
Researchers introduce CauSim, a framework that enables large language models to improve causal reasoning by constructing increasingly complex executable causal simulators. The approach transforms causal reasoning from a scarce-data problem into a scalable supervised learning task, allowing LLMs to generate synthetic training data and demonstrate improved performance across different representations.
