News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
arxiv arxiv-ai | 18d | COMPASS: Routing via Capability-Oriented Model Profiling with Adaptive Search, Selection, and Tool Use | LLM routing aims to select a capable model at low cost, but specialized routers can require updates as models and tasks change. We pro… | ↗Not briefed |
rss marktechpost | 18d | Exa 3.0 Search API Tops Perplexity, Brave and Parallel on 8 Benchmarks | Exa 3.0 is a search API upgrade from Exa. It adds a larger embedding model, Exa Lite at $4 per 1,000 requests, and Dynamic Highlights,… | ↗Not briefed |
arxiv arxiv-ai | 18d | SMART: Semantic Segment Mapping and Assessment of Recognition Transcripts with Contextual Rubrics | Word error rate (WER) assigns equal cost to lexical edits despite their different semantic consequences, while overall semantic scores… | ↗Not briefed |
arxiv arxiv-ai | 18d | Privacy-preserving inference on hosted models: a learned input-embedding transform for frozen Qwen3.8-27B | Every prompt sent to a hosted language model is readable by whoever runs it. We study whether the prompt can remain on the client whil… | ↗Not briefed |
bluesky bluesky-ai | 18d | No, it won’t. | No, it won’t. | ↗Not briefed |
arxiv arxiv-ai | 18d | RAG-NIMBF: A Non-Invasive, Modular Benchmarking Framework for Automated RAG Configuration Analysis | Retrieval-Augmented Generation (RAG) has become the dominant paradigm for grounding the outputs of large language models (LLM) in exte… | ↗Not briefed |
arxiv arxiv-ai | 18d | ProVe: Proposer-Verifier Decoding | Speculative decoding accelerates autoregressive (AR) language models by letting a small model draft tokens serially and a large model… | ↗Not briefed |
bluesky bluesky-ai | 18d | Had a quick read of the paper, though I wouldn’t have the expertise to understand it all, it seems to boil down to ‘If you look older than … | Had a quick read of the paper, though I wouldn’t have the expertise to understand it all, it seems to boil down to ‘If you look older… | ↗Not briefed |
arxiv arxiv-ai | 18d | LatBridge: Efficient and Stable Model Replacement in Latent-Communication Multi-Agent Systems | Multi-agent systems rely on communication to complete tasks, and latent communication passes continuous representations to reduce text… | ↗Not briefed |
hn hn-ai | 18d | CloudSecure – Open-source, local-first CNAPP suite in pure Rust | — | ↗Not briefed |
arxiv arxiv-ai | 18d | SliMOO: Interpretable Multi-Objective Evolutionary Search for LLM Depth Pruning | As large language models (LLMs) grow more powerful, their scale makes them costly to serve. Depth pruning offers a practical route to… | ↗Not briefed |
hn hn-ai | 18d | ZXCode – VS Code programming environment for ZX Spectrum | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Certified Adaptive List Decoding for Error-Correcting Codes | Maximum-likelihood (ML) decoding of linear block codes is NP-hard, so neural decoders are evaluated by how closely they approach it. W… | ↗Not briefed |
bluesky bluesky-ai | 18d | 1) it’s not an LLM 2) I too can look at a photo of someone who looks older than their age because of smoking, drinking, substance abuse, po… | 1) it’s not an LLM 2) I too can look at a photo of someone who looks older than their age because of smoking, drinking, substance abus… | ↗Not briefed |
rss techmeme | 18d | Ant-backed Mynt, which runs mobile wallet app GCash, rose as much as 2.4% in Manila after raising $841M in the Philippines' largest-ever IPO, valuing it at ~$7B (Kimberley Kao/Wall Street Journal) | Kimberley Kao / Wall Street Journal: Ant-backed Mynt, which runs mobile wallet app GCash, rose as much as 2.4% in Manila after raising… | ↗Not briefed |
arxiv arxiv-ai | 18d | Toward Agentic AI for the 6G Physical Layer: From Algorithm Discovery to Verified Baseband Silicon | 6G will be designed faster than it can be built: AI-driven search already proposes physical-layer algorithms at machine speed, while c… | ↗Not briefed |
arxiv arxiv-ai | 18d | TerraNostra-Bench: Benchmarking Spatial Downscaling and Temporal Rollout for Global 1-km Land-Use Reconstruction | Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, partly due to imperfect land surface re… | ↗Not briefed |
arxiv arxiv-ai | 18d | Are Linguistic and Arithmetic Circuits Dissociated in LLMs? | Large language models (LLMs) exhibit both knowledge of language, rendered by, for example, grammatical judgment ability, and abilities… | ↗Not briefed |
arxiv arxiv-ai | 18d | Foundations of the Scaling Law | Empirical scaling laws relate validation loss to model size, training exposure and compute, providing a quantitative basis for allocat… | ↗Not briefed |
arxiv arxiv-ai | 18d | SLM-SS: Token-Based Speech Separation with Speech Language Models | Speech separation (SS) has witnessed substantial progress in recent years with the advent of neural network-based methods, achieving n… | ↗Not briefed |
arxiv arxiv-ai | 18d | RELAY: Scaling Historical Build Restoration through Rule Reuse and Agentic Repair | Vulnerability benchmarks and software engineering research depend on rebuilding historical software. Yet builds rely on external resou… | ↗Not briefed |
bluesky bluesky-ai | 18d | Docs then have to expend tremendous effort justifying why X isn’t appropriate. I know some who have had patients use LLMs during consulta… | Docs then have to expend tremendous effort justifying why X isn’t appropriate. I know some who have had patients use LLMs during con… | ↗Not briefed |
arxiv arxiv-ai | 18d | GRADUS: Discovering and Recovering Agent Lessons from the Environment's Latent Syllabus | Training LLM agents for policy-constrained tool environments, such as an airline backend with booking, refund, and cancellation polici… | ↗Not briefed |
rss marktechpost | 18d | Cloudflare Releases Clairic, a System One Extraction Model on Workers AI | Cloudflare has previewed Clairic, a System One extraction model on Workers AI that returns typed values with confidence scores in one… | ↗Not briefed |
hn hn-ai | 18d | Anyone Used the OAI Batch API or Azure OAI Batch API for Bulk Data Processing? | — | ↗Not briefed |
arxiv arxiv-ai | 18d | RITE: Evaluating End-to-End Resilience to Silent Tool Errors in E-Commerce Agentic Systems | Tool-using LLM agents increasingly answer e-commerce queries through multi-step tool interactions. The expected output of a tool call… | ↗Not briefed |
hn hn-ai | 18d | Will developing nations be able to handle AI agents? | — | ↗Not briefed |
hn hn-ai | 18d | Clippy, the Office Assistant, on the macOS Desktop | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Radar-Based 3D Human Pose Estimation via Motion-Aware Pose Initialization and Observation-Guided Refinement | 3D Human Pose Estimation (3D HPE) is used in indoor environments, such as smart homes, to analyze human movement and monitor human sta… | ↗Not briefed |
arxiv arxiv-ai | 18d | How Much Do $\ell_1$ Attacks Need Concentrated Perturbations? | A global $\ell_1$ budget limits how much an adversary may change an image, not how that change is distributed. We ask how much of an $… | ↗Not briefed |
hn hn-ai | 18d | Pdd-kit – Product-Driven Development for coding agents | — | ↗Not briefed |
hn hn-ai | 18d | Show HN: React-jsonschema-UI – JSON Schema forms for React Native and the web | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Engram Memory Network: Brain-Inspired Prototype Explanations | Artificial intelligence systems are increasingly deployed in high-stakes decision-making. In response, prototype-based models in expla… | ↗Not briefed |
bluesky bluesky-ai | 18d | Folks looking to Anthropic for advice on code reviews need to know where that advice has lead for them. Thankfully, they're giving us a liv… | Folks looking to Anthropic for advice on code reviews need to know where that advice has lead for them. Thankfully, they're giving us… | ↗Not briefed |
bluesky bluesky-ai | 18d | eBay users! Don't ignore today's email, even though it looks like a boring standard update to the User Privacy Notice. They want to use yo… | eBay users! Don't ignore today's email, even though it looks like a boring standard update to the User Privacy Notice. They want to u… | ↗Not briefed |
arxiv arxiv-ai | 18d | Foundation-Style Pretraining for Scalable and Transferable Distributed PV Forecasting | The rapid growth of distributed photovoltaic (PV) generation is increasing the demand for accurate forecasting across large numbers of… | ↗Not briefed |
arxiv arxiv-ai | 18d | CrystalAR: Autoregressive Modeling for Symmetric Crystal Generation | Deep generative modeling has achieved remarkable progress in crystal generation. Given the rich symmetries inherent in crystalline str… | ↗Not briefed |
arxiv arxiv-ai | 18d | AC/DC on a Budget - Alternating Sparse Phases | State-of-the-art neural network sparsification relies on alternating dense and sparse training phases. While computationally expensive… | ↗Not briefed |
arxiv arxiv-ai | 18d | Give the Model a Finger: Developmentally Aligned Evaluation of Counting in Vision-Language Models | Vision-language models (VLMs) are exact on small sets and drift into estimation on larger ones, a profile reminiscent of children befo… | ↗Not briefed |
arxiv arxiv-ai | 18d | Detection of Memorization in Diffusion Models Through Thermodynamic Singularity | In diffusion models, proper distributional learning entails reversing a diffusion process over underlying density functions, whereas a… | ↗Not briefed |
hn hn-ai | 18d | Should LLMs write in mainstream languages? | — | ↗Not briefed |
hn hn-ai | 18d | Instagram without reels, across updates | — | ↗Not briefed |
hn hn-ai | 18d | Physicists prove a long-suspected quantum speed limit for thermalization | — | ↗Not briefed |
hn hn-ai | 18d | Revolut down – Global ATM withdrawals and bank transfers failing | Revolut has been down almost an hour, with ATM withdrawals across multiple countries failing, bank transfers not working, and various… | ↗Not briefed |
hn hn-ai | 18d | Marble: Exact sequence parallelism for packed multimodal masks | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Constant Step Sizes for Muon-Type Spectral Optimizers on Strongly Convex Quadratics | Muon and related optimizers update a weight matrix with a spectrally transformed gradient. In this work, we study such spectral optimi… | ↗Not briefed |
arxiv arxiv-ai | 18d | S$^2$-MAS: A Society-Scale Multi-Agent Simulation Framework with 500 Million Household Persona Agents | Simulating economic responses at society scale requires agents whose decisions reflect household circumstances and the effects of earl… | ↗Not briefed |
arxiv arxiv-ai | 18d | Hearsay or Fact? How Framing and Elicitation Shape LLM Probability Updates🔧 Elicit | Large language models (LLMs) are increasingly used as judgment assistants and as stand-in participants in psychology experiments, yet… | ↗Not briefed |
bluesky bluesky-ai | 18d | Draft bill tied to Anthropic would let courts halt AI models for 45 days #AI #ArtificialIntelligence #TechPolicy #MachineLearning #LegalTech | Draft bill tied to Anthropic would let courts halt AI models for 45 days #AI #ArtificialIntelligence #TechPolicy #MachineLearning #Leg… | ↗Not briefed |
bluesky bluesky-ai | 18d | Draft bill tied to Anthropic would let courts halt AI models for 45 days #AI #ArtificialIntelligence #TechPolicy #MachineLearning #LegalTech | Draft bill tied to Anthropic would let courts halt AI models for 45 days #AI #ArtificialIntelligence #TechPolicy #MachineLearning #Leg… | ↗Not briefed |