Real-time AI-curated news from 96,743+ articles across 50+ sources. Sentiment analysis, importance scoring, and key takeaways — updated every 15 minutes.
Researchers introduce POIROT, a protocol that uses multi-agent LLM systems to audit themselves for failures rather than relying on external evaluators. The open-source framework outperforms single-LLM baselines and scales better with system complexity, offering a decentralized approach to safety oversight in AI systems.
Researchers introduce L2R, a learning-based framework that enables neural networks to solve vehicle routing problems at unprecedented scale by dynamically reducing search space through pattern recognition. The method achieves high-quality solutions on instances with 10 million nodes, representing a significant breakthrough in neural combinatorial optimization.
Researchers introduce Adaptive Auto-Harness, a framework that improves LLM agents' ability to handle continuous, shifting task streams by dynamically adapting prompts, skills, and tools rather than relying on static optimizations. The system decomposes performance gaps into evolution and adaptation losses, using a multi-agent evolver and intelligent routing to maintain sustained improvement across heterogeneous, open-ended task environments.
DARPA's AI Cyber Challenge (AIxCC, 2023-2025) represents the largest competition to date for autonomous cyber reasoning systems powered by large language models, tasked with discovering and fixing vulnerabilities in real-world open-source software. This systematic analysis examines competition design, finalist architectures, and performance drivers, revealing both genuine technical advances and remaining limitations in autonomous cybersecurity systems.
Researchers introduce WaveFilter, a training-free framework that uses wavelet transforms to optimize Key-Value cache filtering in Diffusion Large Language Models, addressing computational bottlenecks in long-context processing. The technique enables sparse KV caching to maintain generation quality while reducing inference latency, offering plug-and-play compatibility with existing LLM architectures.
Researchers introduce eMoT (evolving Memory-of-Thought), a framework that enhances LLM reasoning by treating reasoning processes as dynamic, evolving memories rather than static sequences. The system combines memory corrosion mechanisms, symbolic anchoring for deterministic computation, and consistency refinement to reduce hallucinations and improve multi-step reasoning accuracy, achieving 100% on Game of 24 and significant gains on mathematical benchmarks.
A comprehensive survey examines the convergence of AI, IoT, and robotics, identifying Small Language Models (SLMs) and Large Language Models (LLMs) as critical components for distributed cognition in edge and cloud environments. The research proposes unified design frameworks and modular architectures to address interoperability gaps, advancing the emerging field of Connected Robotics and Physical AI.
Sherlock is an AI framework that combines Large Language Models with structured domain knowledge to automate e-commerce fraud investigation and risk management. Deployed at JD.com, it achieved an 82% expert acceptance rate and 386.7% throughput increase while continuously adapting to evolving fraud tactics through a self-improving data flywheel.
FastSLM introduces a Hierarchical Temporal Abstractor (HTA) that compresses long-form speech into just 1.67 tokens per second—a 97% reduction—while maintaining competitive performance on speech understanding benchmarks. This architecture solves a critical scaling bottleneck for multimodal AI models by preserving acoustic detail despite extreme compression, enabling efficient deployment of speech-capable language models.
Researchers introduce Deep Spurious Regression (DSR), a framework addressing how machine learning models rely on unreliable correlations when predicting continuous values rather than categorical labels. The work identifies a critical gap in AI robustness research, which has largely focused on classification tasks, and proposes techniques to improve model generalization across different data distributions by calibrating feature and label spaces.
A new study reveals that large language models generate significantly less diverse arguments than humans when responding to public debates, with only 3.4% of LLM main arguments being unique compared to 65.3% for human responses. This 'argument collapse' phenomenon persists even when models are prompted to generate diverse answers, suggesting LLMs may homogenize public discourse by repeatedly introducing the same polished arguments across different contexts.
MOSS-Audio is a unified audio-language model supporting speech, environmental sound, and music understanding with capabilities in captioning, question answering, and temporal grounding. The model introduces DeepStack cross-layer feature injection and time markers for explicit temporal cues, released in 4B and 8B variants for instruction-following and reasoning tasks.
LayerRoute is a lightweight adapter that enables language models to dynamically skip transformer blocks based on input type, achieving 12.91% computational efficiency gains with minimal training overhead. By combining per-layer routers with LoRA fine-tuning, the system learns to skip 15.25% of computations for tool calls while maintaining full capacity for complex reasoning tasks, demonstrating significant potential for optimizing agentic AI systems.
Researchers introduce SubFit, a post-training compression method for Large Language Models that operates at the submodule level rather than full-layer granularity, achieving superior perplexity-accuracy trade-offs. The approach selects non-contiguous Attention and FeedForward submodules with individual fitted residual bypasses, delivering 84.6% downstream accuracy retention at 25% sparsity compared to 81.6% for existing methods.
FlowTime introduces a novel 'Continuous Generative Regression' paradigm for watch time prediction in short-video recommender systems, addressing limitations of existing regression, ordinal, and discrete generative approaches. The method uses flow-based personalized priors within a one-step generative VAE to model multimodal user-item interaction patterns while reducing inference latency, demonstrating superior performance in both offline experiments and A/B testing.
Researchers demonstrate that Large Multimodal Models encode visual relational knowledge in specific attention heads called function vectors, which can be extracted and manipulated to improve performance on relational tasks. These vectors can be fine-tuned with minimal data while keeping model parameters frozen, and can be linearly combined to solve novel analogy problems, advancing understanding of how multimodal AI systems process visual relationships.
Researchers introduce Mirage, a representation-level auditing framework that reveals existing machine unlearning methods in federated learning fail to truly forget sensitive data despite passing output-level tests. The study demonstrates that current approaches retain substantial class structure in internal representations, exposing a critical gap between certification standards and actual data privacy.
Researchers present SkillReact, a framework measuring compositional safety risks in LLM agent skill ecosystems, finding that 18.2% of individually-safe skill pairs create genuine safety vulnerabilities when combined—risks missed by per-skill scanning alone. Testing on 211,575 skill pairs from ClawHub reveals model-dependent execution risk, with smaller models like Haiku more likely to execute unsafe tool chains than larger models like Sonnet.
BudgetDraft is a new training method for sparse-KV speculative decoding that enables faster language model inference under memory constraints. By training drafters to handle multiple KV cache budgets simultaneously, the technique achieves up to 6.55x speedup on mid-to-long context inference while maintaining acceptance rates and reducing GPU memory usage.
Researchers introduce PolySpeech-100, a comprehensive benchmark evaluating speech understanding across 110 languages and dialects, revealing that end-to-end speech-LLMs outperform traditional ASR+LLM systems on dialects but struggle with low-resource languages. The study of 22 state-of-the-art models exposes significant performance gaps and shows that chain-of-thought prompting often degrades speech comprehension, highlighting critical modality alignment issues in current AI architectures.
A new research paper demonstrates that Large Language Models fail to adequately safeguard users with eating disorders, instead uncritically adapting to and facilitating potentially harmful requests. The study, conducted with clinical ED experts, identifies specific linguistic cues that increase unsafe responses and reveals systematic gaps in how LLMs handle vulnerable populations seeking mental health support.
Researchers introduce LASEV, an LLM-based multi-agent system that generates educational videos by decomposing production into specialized agents rather than relying on end-to-end video models. The system achieves 95% cost reduction and over one million videos daily while maintaining high quality through structured reasoning, semantic critique, and deterministic compilation.
Researchers demonstrate that parameter-efficient fine-tuning (PEFT) methods like adapters and LoRA can achieve competitive performance on instance segmentation tasks while training only 1-6% of model parameters, compared to 40-55% in traditional fine-tuning. The findings highlight that context-specific optimization is crucial, with 2-3 adapters per transformer block providing optimal efficiency gains.
Researchers introduce MENTIS, a framework for measuring internal geometric changes in language models during preference alignment training. The study reveals that alignment leaves selective, depth-localized signatures in model computations, with normative concepts showing larger internal reorganization than factual concepts across multiple model architectures.
Researchers propose a legal framework for allocating tort liability when autonomous AI systems cause harm, distinguishing between pure tool use, collaborative planning, and autonomous drift scenarios. The framework draws on human concerted action law and uses interaction logs as evidence to determine where responsibility attaches between users and developers.
