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 | 1mo | K2P: Label-Free Knowledge to Prompt Distillation | Knowledge distillation can transfer reasoning from stronger teachers to frozen students through reusable prompts, but avoiding weight… | ↗Not briefed |
rss techmeme | 1mo | A profile of Mariano-Florentino Cuéllar, Anthropic's first global affairs chief, who had helped shape California's first AI safety law in 2025 (Brandon Pho/Mission Local) | Brandon Pho / Mission Local: A profile of Mariano-Florentino Cuéllar, Anthropic's first global affairs chief, who had helped shape Cal… | ↗Not briefed |
arxiv arxiv-ai | 1mo | FFASR: Benchmarking Far-Field Automatic Speech Recognition using High-Fidelity Simulated RIRs | Far-field automatic speech recognition(ASR) degrades under reverberation, noise, and talker motion, yet the benchmarks that drive mode… | ↗Not briefed |
hn hn-ai | 1mo | MPH Drone RTOS | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | the 12axes test at 12axes.vercel.app is vibe coded. the github page github.com/RomanCypherp... has a folder called ".claude" and Vercel is … | the 12axes test at 12axes.vercel.app is vibe coded. the github page github.com/RomanCypherp... has a folder called ".claude" and Verce… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Unmerge: Efficient Machine Unlearning via Task Arithmetic | Approximate machine unlearning seeks to remove the influence of a forget set from a trained model without full retraining. Existing gr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Continuous Neural Representation of Stochastic Hybrid Systems | A stochastic hybrid system (SHS) is governed by a stochastic differential equation (SDE) describing the continuous dynamics and a Mark… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Don't start calling it Super Intelligence, please. That's so stupid. A pure propaganda scheme and a lie. Don't even call it AI. Because it… | Don't start calling it Super Intelligence, please. That's so stupid. A pure propaganda scheme and a lie. Don't even call it AI. Becau… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Consistent Plan-Act for Long-Horizon Agentic Tasks🔧 GPT-5 | Long-horizon agentic tasks demand strong reasoning and efficient execution across successive interactions with dynamic environments. A… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VERA: Verifiable Feasibility Representations with Counterfactual Credit for Constrained Multi-Agent Control | Constrained multi-agent control requires more than predicting rewarding actions: an action can cease to be executable as contact windo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VOSSA: Voiceprint Optimization for Streaming Speech Architectures | Real-time voice conversion (VC) systems commonly rely on pretrained speaker embeddings from automatic speaker verification (ASV) model… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: How do you handle hitting AI coding agent usage limits? | One topic I've been researching lately is the best way to handle switching to another agent when I hit a usage limit on one of th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Right Answers, Costly Models: The Efficiency Gap in LLM-based Optimization Modeling | Optimization modeling formulates real-world decision problems as mathematical programs that solvers can use to find optimal decisions.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If you issue a public apology and offer pro bono IT services, you can not only get away with cybercrime but also monetize it through instit… | If you issue a public apology and offer pro bono IT services, you can not only get away with cybercrime but also monetize it through i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | STRATA: Self-Learning Through Role-Aligned Tiered Agents for Real-Time Strategy Games | Real-time strategy (RTS) games require agents to coordinate economic development, production and construction, base defense, unit orga… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sharp Statistical Rates for Asynchronous TD Learning with Markovian Data | We study the last iterate of standard tabular temporal-difference (TD) learning from a single trajectory of a finite Markov reward pro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | the best part of this is they clearly used an LLM to check it and it just hallucinated out the answer it thought they wanted to hear | the best part of this is they clearly used an LLM to check it and it just hallucinated out the answer it thought they wanted to hear | ↗Not briefed |
arxiv arxiv-ai | 1mo | Does Learning Protein Folding Generalize to Broader Reasoning? | Large language models rely heavily on human text, which often conveys surface answers rather than the spatial and structural logic beh… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Audio Token Attention Is Predictable Before the Language Model Runs | A large audio language model (LALM) turns a minute of speech into 750-1,500 tokens and prefills every one. Image-token pruning often c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | On Parameters of Nonlinear Scalar Dynamics from Video: Invariants, Calibration, and Identifiability | Physical parameter estimation from video aims to recover the parameters of a known family of governing dynamical equations from pixel… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reasoning Externalization for Faithful Large Language Model Narratives of Stock Return Predictions | In finance, interpreting machine learning predictions is essential, yet the numerical outputs of explainable AI can be difficult for n… | ↗Not briefed |
hn hn-ai | 1mo | PSSA: A non-transformer language model written from scratch in Rust | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Talk2Agent: Benchmarking Voice Interfaces for Text Agents | Large language model (LLM) computer-use agents are typically evaluated with clean written instructions, despite speech being an increa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Context Changes: Understanding Update Failures in LLMs | As preferences, goals, and facts change, LLM agents must use the current state while earlier versions remain in context. Yet they can… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GeoNest: Learning to Select Failure-Aware Neighborhoods for the Irregular Knapsack Problem in a Circular Container | The two-dimensional irregular knapsack problem in a fixed circular container is an important combinatorial optimization problem for ma… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Multi-Modal Planning with Reward-Guided Preference Optimization for Autonomous Driving🔧 Modal | Safe and efficient trajectory planning is essential in autonomous driving. However, existing end-to-end approaches often fall short in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TRACE: Target-Aware Retrieval, Attributed Evidence, and Contract-Constrained Extraction for LitTraceQA | Finding a relevant paper is not the same as producing a verifiable answer from it. LitTraceQA requires canonical paper identifiers, ex… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimal Design for Active Preference Learning with Biased LLM Judges | Learning from human preferences is central to large language model (LLM) alignment, but human preference annotation is costly. Active… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Online Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization | Vision-Language-Action (VLA) models based on generative frameworks, such as Flow Matching, have recently achieved impressive performan… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Mitigating the Length-Scaling Tax with Online Distillation | Length scaling during reinforcement-learning (RL) post-training is often viewed as a sign of improved reasoning ability, especially on… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Visualizing Distribution Coverage in Generative Diffusion Models | Diffusion distillation is widely adopted to accelerate sampling, and the resulting few-step models are broadly believed to match or ev… | ↗Not briefed |
hn hn-ai | 1mo | Critical LiteLLM Vulnerability | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Where MLLMs Fail and Why: Causal Task Decomposition for Capability Failure Diagnosis | End-to-end accuracy on compositional tasks records how often MLLMs fail, but cannot distinguish whether a failure reflects an intrinsi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OpenJev-RLCD: A Working RLCD Implementation | Decision models such as Jev answer questions with probabilities, which are only useful if they are calibrated. Open-source reproductio… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics | Rubric-based reinforcement learning (Rubric-RL) trains language models where no verifier exists. A judge checks each criterion of a ru… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learn-Then-Differentiate Gradient Estimation | Learn-then-differentiate (LTD) estimates gradients by fitting a model to simulation outputs and differentiating it. We develop a unifi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAIって書いてるの見ると毎回OppAIに空目してしまう。2文字しか違わないからほぼそうと言えるしいいか…… | OpenAIって書いてるの見ると毎回OppAIに空目してしまう。2文字しか違わないからほぼそうと言えるしいいか…… | ↗Not briefed |
hn hn-ai | 1mo | DeepGEMM-Ascend: Huawei Ascend NPUs Matrix Multiplication Kernel Library | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | cute that ai assumes i need an iso. no, little buddy; original media. | cute that ai assumes i need an iso. no, little buddy; original media. | ↗Not briefed |
hn hn-ai | 1mo | RSS Feeds for Last.fm | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | scTrilemma: Balancing Identity, Invariance, and Fidelity in Single-Cell Representation Learning | Single-cell RNA-seq representation learning is fundamentally label-free: cell identities, states, and contexts are not fixed training… | ↗Not briefed |
hn hn-ai | 1mo | Why Didn't GitHub Copilot Win? | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Generate live realtime dashboards of your system in seconds | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Average-and Last-Iterate Lower Bounds for Optimistic Matrix Mirror-Prox in Quantum Zero-Sum Games | Optimistic matrix mirror-prox (OMMP) computes $ε$-approximate Nash equilibria in quantum zero-sum games with an $O(1/\varepsilon)$ ave… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimal VC Dimension of Contrastive Learning with Margin | Contrastive learning is a successful paradigm for learning $d$-dimensional geometric representations from a collection of ``anchor--po… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ReSCENE: Server-Side Replay for Structural Mitigation of Catastrophic Forgetting in Federated Continual Learning | Federated continual learning must integrate new tasks over time without losing earlier-task knowledge. Most existing methods attach an… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scaling Parameter and Context in Attention: Native Sparse Attention from Mixture-of-Head | Scaling attention parameters can improve language model quality, but retaining full token histories makes additional heads costly at l… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Nothing like unfollowing someone because they retreated news about a new ai model agent... nice. LOVE being an artist in the 2026 vtubing s… | Nothing like unfollowing someone because they retreated news about a new ai model agent... nice. LOVE being an artist in the 2026 vtub… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Forging LLM Authorship Fingerprints with Targeted Rewriting | Model-attribution classifiers can often identify which language model produced a text, making model-specific writing patterns a signal… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly: I'm really glad China went with an open weights LLMs strategy and at least currently doesn't plan to get away from that strategy … | Honestly: I'm really glad China went with an open weights LLMs strategy and at least currently doesn't plan to get away from that stra… | ↗Not briefed |