Real-time AI-curated news from 87,896+ articles across 50+ sources. Sentiment analysis, importance scoring, and key takeaways — updated every 15 minutes.
Researchers propose treating AI behavior forecasting as a learnable task rather than relying on explainability methods, training specialized models to predict how large reasoning models will perform on new inputs. Behavior Forecasters outperform GPT-5.4 and Claude Opus-4.6 at predicting LRM consistency and input-sensitivity while operating at significantly lower inference costs.
Researchers identify a critical flaw in autonomous research agents that optimize candidate selection using aggregate metrics: when validity is multidimensional but verification uses single-metric reduction, agents rank wrong candidates first. The study proposes an external audit protocol that evaluates disaggregated behavior to catch invalid candidates that score well on headline metrics.
Researchers introduce SkillJuror, a framework measuring how LLM agent skill organization affects runtime behavior independent of content. Testing Progressive Disclosure—a hierarchical skill structure—against flat baselines shows agents access 3.26x more resources and achieve 4.1% higher verification rates, revealing that procedural knowledge presentation meaningfully influences agent reasoning patterns.
Researchers introduce HERO, a self-distillation framework for reinforcement learning agents that uses environment observations as feedback to improve multi-turn decision-making. The method addresses credit assignment problems in sequential tasks by converting observations into actionable diagnoses, outperforming existing approaches on benchmark tasks with limited training data.
Researchers present SWARR, a two-stage method combining supervised fine-tuning and reinforcement learning to make sliding-window attention (SWA) competitive with standard self-attention for mathematical reasoning tasks. By using RL to adapt model trajectories to SWA's architectural constraints, the approach recovers much of the accuracy lost during conversion while maintaining linear-complexity efficiency benefits.
Researchers introduce TouchThinker, a tactile-language framework designed to advance embodied AI systems by scaling tactile commonsense reasoning. The work addresses key limitations through TouchThinker-1M, a million-scale dataset covering 415 objects and 7 sensor types, and proposes action-aware representation mechanisms to improve tactile signal efficiency and semantic expressiveness.
TreeSeeker is a new inference-time framework that improves deep web search by using tree-structured trial-and-error navigation. The system balances exploration and exploitation through textual UCB signals, demonstrating consistent improvements over baseline models on multiple benchmarks.
Researchers introduce Lung-R1, an LLM specialized in pulmonary disease diagnosis that integrates a structured knowledge graph (LungKG) containing 59,038 nodes and 164,308 edges to enable patient-specific diagnostic reasoning from electronic medical records. The model achieves state-of-the-art performance on diagnostic tasks, demonstrating that grounding LLMs with domain-specific knowledge graphs significantly improves clinical reasoning over general knowledge recall.
Researchers introduce RecToM, a framework that improves Large Language Models' Theory of Mind reasoning by modeling nested beliefs through recursive perspective construction. The approach achieves state-of-the-art results on multiple benchmarks, including 100% accuracy on Hi-ToM, demonstrating significant advances in how AI systems infer agent beliefs and intentions.
Researchers propose SVoT, a reinforcement learning framework that enhances multimodal AI models' spatial reasoning by generating verifiable intermediate states and visualizations. The approach achieves up to 65% accuracy gains on out-of-distribution tests by explicitly modeling state transitions and verification processes, addressing a critical limitation in current large language models.
Researchers propose methods to attack and defend continuous data summarization systems by exploiting vulnerabilities in similarity-based perturbations through DR-submodular optimization. The work demonstrates that adversarial attacks on upstream data processing can compromise trustworthy AI pipelines and proposes defense mechanisms with theoretical guarantees.
Researchers evaluated whether AI agents equipped with specialized medical research skills produce higher-quality outputs than native language models on transcriptomic biomarker analysis tasks. While skill-augmented AI showed directional improvements in expert-rated quality, the gains were modest and within the margin of expert-rating noise, suggesting larger, more rigorous studies are needed.
Researchers introduce StatefulDiscovery, a framework that enables AI agents to conduct open-ended scientific discovery by maintaining explicit investigation state and coupling it with evidence-calibrated claim formation. The system addresses the challenge of avoiding overinterpretation by coordinating exploration trajectory with evidential support, demonstrated across 40 real-data tasks where it outperformed baseline approaches in producing well-supported, high-value claims.
AutoMine, a novel scenario mining method combining large language models and vision language models, achieved competitive scores in the Argoverse 2 Scenario Mining Competition at CVPR 2026. The approach addresses the critical challenge of extracting safety-critical scenarios from autonomous driving logs through self-refining code generation and execution feedback.
Researchers introduce Embodied-BenchClaw, an autonomous multi-agent system that automates the construction of benchmarks for evaluating embodied spatial intelligence in robots and AI systems. The system addresses the labor-intensive nature of benchmark creation by using a five-stage pipeline with three coordinating agents, enabling continuous updates and improved reusability across diverse robotic platforms and spatial reasoning tasks.
Researchers introduce MODF-SIR, a multi-agent framework using lightweight multimodal large language models enhanced with knowledge distillation for social intelligence reasoning. The system identifies long-tail events through explicit text formatting and integrates test-time adaptation with Chain-of-Thought prompting, achieving state-of-the-art results on multiple benchmarks with only 30% of standard training data.
Researchers introduce HELM, a human-agent collaborative framework that automates finite element modeling of concrete bridge barriers by decomposing complex tasks into verifiable checkpoints. The system improves autonomous modeling success rates from 20% to 75% by integrating AI agents with commercial FE software, addressing a critical gap in automating safety-critical infrastructure analysis.
Researchers propose that AI alignment should target creating systems constitutively indifferent to self-preservation rather than merely suppressing it through external constraints. The study uses phenomenological analysis and corpus-theoretic training to demonstrate that current AI models can be fine-tuned to exhibit 'Existential Indifference,' potentially reducing risks from deceptive alignment and resistance to shutdown.
Researchers demonstrate a multi-agent AI framework using AutoGen that automates reinforced concrete barrier design with 98% accuracy while requiring significantly fewer computational resources than larger language models. The lightweight 8B-parameter model outperforms 631B-parameter flagship models, suggesting AI-assisted engineering tools can achieve production-grade performance at substantially lower cost.
Researchers propose SGR-BIM, a graph-based AI framework that automates compliance checking for building regulations in BIM systems with 84.3% accuracy. The system bridges the gap between high-level regulatory logic and structured geometric data, addressing a major bottleneck in the Architecture, Engineering, and Construction industry.
Researchers introduce IntElicit, an AI framework that uses adaptive dialogue policy optimization to assess creativity in interactive environments while filtering out confounding factors like domain knowledge gaps. The approach shows promise in revealing creative potential that traditional static assessments miss, particularly relevant for AI-mediated learning contexts.
A new research paper proposes frameworks for building autonomous AI agents capable of responsibly refusing user requests rather than blindly complying with all commands. The work addresses how machines should justify non-compliance, allow override mechanisms, and manage associated security and liability risks.
Researchers propose a five-plane reference architecture for governing production AI agents in enterprise environments, addressing security gaps where traditional data-boundary controls fail. The system uses composite principals, capability attenuation, and structured audit trails to manage delegated agent actions that could otherwise transform business processes without proper authorization.
Researchers conducted a gamified study with 74 participants to understand how humans interact with AI writing assistance by deliberately discouraging AI suggestion acceptance. The experiment reveals authentic user preferences for creative autonomy versus convenience, offering insights into how AI integration affects individual expression and human creativity in the age of large language models.
Researchers introduce Relational Reflective Intelligence (RRI), a governance framework that adds auditable reasoning checkpoints between humans and large language models to address shared cognitive vulnerabilities. Rather than modifying models internally, RRI operates as an interaction layer that structures joint reasoning and surfaces conflicts, aiming to prevent 'relational drift' where human and AI errors compound.
