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 |
|---|---|---|---|---|
bluesky bluesky-ai | 21h | Ender's Game was a documentary | Ender's Game was a documentary | ↗Not briefed |
arxiv arxiv-ai | 21h | HiSA: Sparsifying the Hidden Dimension of RoPE Attention in LLMs | Attention in large language models (LLMs) scores every query against every key with a full dot product. We investigate whether this do… | ↗Not briefed |
bluesky bluesky-ai | 21h | The overlap of retail investors cheering for the money they're making on the Anthropic IPO and those advocating for no new datacenters? | The overlap of retail investors cheering for the money they're making on the Anthropic IPO and those advocating for no new datacenters? | ↗Not briefed |
bluesky bluesky-ai | 21h | Best prep I got for LLMs was a Linear Computation summer course in my undergrad back in the mid-00s, where we dug into highly-parallelized … | Best prep I got for LLMs was a Linear Computation summer course in my undergrad back in the mid-00s, where we dug into highly-parallel… | ↗Not briefed |
bluesky bluesky-ai | 21h | Oh no, wut | Oh no, wut | ↗Not briefed |
bluesky bluesky-ai | 21h | I don't mind people using AI assistance to contribute to projects I maintain. I do mind people fucking doing drive-by PRs where they move o… | I don't mind people using AI assistance to contribute to projects I maintain. I do mind people fucking doing drive-by PRs where they m… | ↗Not briefed |
arxiv arxiv-ai | 21h | How Much Should an Agent Remember? Deployment Lessons from Self-Distilled Memory for LLM Agents | LLM agents on long-horizon tasks fail in recurring, procedural ways. They mis-paginate an API, resolve the wrong entity, or answer whe… | ↗Not briefed |
arxiv arxiv-ai | 21h | Learning from Context: Leave-One-Out Prediction in Jump Diffusions | Continuous diffusion models generate discrete data through stochastic dynamics in an embedding space. We consider jump diffusions that… | ↗Not briefed |
bluesky bluesky-ai | 21h | Conversely, our creche sends out newsletters that I know for a fact are written by a person but whose tone and vibe make it seem like they’… | Conversely, our creche sends out newsletters that I know for a fact are written by a person but whose tone and vibe make it seem like… | ↗Not briefed |
arxiv arxiv-ai | 21h | Beam Model Technical Report | Beam is a sparse Mixture-of-Experts (MoE) model with 501.3B total parameters and 23.4B active parameters, specifically built and optim… | ↗Not briefed |
bluesky bluesky-ai | 21h | Little OpenAI logo on ‘em and everything | Little OpenAI logo on ‘em and everything | ↗Not briefed |
hn hn-ai | 21h | Inside PyPy: First Contact with RPython | — | ↗Not briefed |
bluesky bluesky-ai | 21h | YES. that's why I've been trying hard on my everyday life to call what most call "AI" as "LLM"s. If the president can rebrand it as "SI" we… | YES. that's why I've been trying hard on my everyday life to call what most call "AI" as "LLM"s. If the president can rebrand it as "S… | ↗Not briefed |
hn hn-ai | 21h | From Static to Current | — | ↗Not briefed |
arxiv arxiv-ai | 21h | Mamba Flow Matching Neural Processes: Linear-Time Inference for Irregularly Observed Spatial Fields | Interpolating a spatial field, such as daily maximum temperature, from sparse, irregularly located observations recurs across the envi… | ↗Not briefed |
bluesky bluesky-ai | 21h | This is part of why I reject the "early days" framing for LLM development. I think we're quite far along by any historical timeline for a t… | This is part of why I reject the "early days" framing for LLM development. I think we're quite far along by any historical timeline fo… | ↗Not briefed |
arxiv arxiv-ai | 21h | TRCF: Temporal-Reference Complementary Forecasting for Irregular Multivariate Time Series | Irregular multivariate time series forecasting supports healthcare monitoring, environmental analysis, and activity recognition. Chann… | ↗Not briefed |
bluesky bluesky-ai | 21h | Strange! Maybe they should supplement their plans with some cheap deepseek etc as backup. It's amazing how much you can do even paying per-… | Strange! Maybe they should supplement their plans with some cheap deepseek etc as backup. It's amazing how much you can do even paying… | ↗Not briefed |
arxiv arxiv-ai | 21h | Perfumer: A Large Language Model-based Agentic Framework Accelerating Structural Relaxation and Electronic Band Prediction for First-Principle Calculation | First-principles structural relaxation and band calculations require substantial computation and expertise. Here we present Perfumer,… | ↗Not briefed |
hn hn-ai | 21h | Show HN: A self-hosted OSS registry for Hugging Face models and datasets🔧 Hugging Face | — | ↗Not briefed |
bluesky bluesky-ai | 21h | Is it wrong to work on projects that will cause human extinction? No, it's the humans who are wrong. But seriously, it's weird they think… | Is it wrong to work on projects that will cause human extinction? No, it's the humans who are wrong. But seriously, it's weird they… | ↗Not briefed |
bluesky bluesky-ai | 21h | Yeah. The point of academia is that we develop communities of scholars with expertise, and they can evaluate and critique each others' work… | Yeah. The point of academia is that we develop communities of scholars with expertise, and they can evaluate and critique each others'… | ↗Not briefed |
bluesky bluesky-ai | 21h | Anthropic raised $107.3 billion in the biggest IPO in history. Here's what to know about the company and its debut on the public markets | Anthropic raised $107.3 billion in the biggest IPO in history. Here's what to know about the company and its debut on the public marke… | ↗Not briefed |
arxiv arxiv-ai | 21h | Context as a Resource | Managing concurrent tasks across separate agent sessions quickly exhausts human attention, motivating always-on personal agents behind… | ↗Not briefed |
rss techcrunch-ai | 21h | At Nasdaq opening bell, Anthropic CEO recommits to pacing AI development | In his Nasdaq opening bell speech, Dario Amodei reiterated Anthropic's commitment to pacing AI development, regardless of business or… | ↗Not briefed |
bluesky bluesky-ai | 22h | like no, schools should not be buying LLM subscriptions for their undergrads, but "just cancel all the contracts" is an insane demand. Ther… | like no, schools should not be buying LLM subscriptions for their undergrads, but "just cancel all the contracts" is an insane demand.… | ↗Not briefed |
bluesky bluesky-ai | 22h | Yes, this is the goal. This is the literal point of academic expertise and journal review. Your manuscript will be sent to one of the preem… | Yes, this is the goal. This is the literal point of academic expertise and journal review. Your manuscript will be sent to one of the… | ↗Not briefed |
arxiv arxiv-ai | 22h | DeepQuest: From Knowledge Gaps to Scientific Discovery in Alzheimer's Disease | Automating experiments does not necessarily advance scientific understanding. Research agents must identify meaningful knowledge gaps,… | ↗Not briefed |
arxiv arxiv-ai | 22h | Training LLMs to Follow Reasoning Instructions with Minimal Edits | Large language models increasingly rely on extended reasoning traces, yet the training signals used to shape this behavior remain coar… | ↗Not briefed |
arxiv arxiv-ai | 22h | Sirius-Tiamat: How Small Can a One-Pass Decision Encoder Get Before Its Decisions Change? | A one-pass decision encoder reads a text, a question and several named options, and returns a probability for each option in a single… | ↗Not briefed |
bluesky bluesky-ai | 22h | And the AI affordances that undermine ethical, rigorous claims are hidden inside the model. You can’t interrogate them, sort them, track th… | And the AI affordances that undermine ethical, rigorous claims are hidden inside the model. You can’t interrogate them, sort them, tra… | ↗Not briefed |
arxiv arxiv-ai | 22h | One Framework to Diagnose Them All: A Stratified Benchmark for Bengali LLM Hallucination | Hallucination research in low-resource languages fragments easily: each newly named failure mode prompts its own dataset, spending sca… | ↗Not briefed |
bluesky bluesky-ai | 22h | *ANTHROPIC SHARES INDICATED TO OPEN AT $108 EACH, IPO PRICE $83 | *ANTHROPIC SHARES INDICATED TO OPEN AT $108 EACH, IPO PRICE $83 | ↗Not briefed |
bluesky bluesky-ai | 22h | Jeffries paying Elon and Anthropic/OpenAI for market research on Americans and members of Congress isn’t totally outside his scope as the m… | Jeffries paying Elon and Anthropic/OpenAI for market research on Americans and members of Congress isn’t totally outside his scope as… | ↗Not briefed |
arxiv arxiv-ai | 22h | ClosureBench: A Benchmark for Reasoning under Closure Assumptions | AI agents built on large language models (LLMs) often answer questions from incomplete records. Missing information might be interpret… | ↗Not briefed |
hn hn-ai | 22h | Show HN: Copilot Instructions Sanity Check | — | ↗Not briefed |
arxiv arxiv-ai | 22h | Online Temporal Rotations for 2 Bit KV-Cache Quantization in Video Diffusion Transformers | Video diffusion transformers reuse key--value (KV) caches across denoising steps, which quickly becomes a bottleneck for long video ge… | ↗Not briefed |
arxiv arxiv-ai | 22h | OceanCoder: A Verification-Guided Large Language Model for Automated Simulation Modelling in Ocean Engineering | Numerical simulation is important in ocean engineering, but model construction requires expert knowledge. Large language models suppor… | ↗Not briefed |
arxiv arxiv-ai | 22h | Rigel-Ladon: An Option-Count Law for a One-Pass Decision Encoder, Tested on Held-Out Option Counts and Held-Out Schemas | Laya is a small neural network that reads a text, a question and K written options, and returns one probability per option in a single… | ↗Not briefed |
bluesky bluesky-ai | 22h | lmao suck it spacexai | lmao suck it spacexai | ↗Not briefed |
bluesky bluesky-ai | 22h | “The city of Tacoma no longer has a city reporter,” Sherred said. “What’s the purpose of a local paper if you have no one covering your loc… | “The city of Tacoma no longer has a city reporter,” Sherred said. “What’s the purpose of a local paper if you have no one covering you… | ↗Not briefed |
bluesky bluesky-ai | 22h | … et quoi qu'on dise, la traduction neuronale et les LLM font énormément d'erreurs et de pas terrible. On le voit tous les jours, même quan… | … et quoi qu'on dise, la traduction neuronale et les LLM font énormément d'erreurs et de pas terrible. On le voit tous les jours, même… | ↗Not briefed |
bluesky bluesky-ai | 22h | c-suite guy in a meeting just said that AI is so useful "just like a calculator" and I'm starting to think people who say that just have ne… | c-suite guy in a meeting just said that AI is so useful "just like a calculator" and I'm starting to think people who say that just ha… | ↗Not briefed |
arxiv arxiv-ai | 22h | Conformal Selection for Hierarchical Streaming Data | Many real-world tasks involve making sequential decisions about whether to select candidates before their responses of interest are re… | ↗Not briefed |
hn hn-ai | 22h | The Adversary – Open-source red-team wireless firmware for ESP32 | — | ↗Not briefed |
arxiv arxiv-ai | 22h | Predicting Surface Displacement Fields of Laser Powder Bed Fusion Parts Using STL-Derived Graphs and Part-Level Descriptors | Predicting part distortion before fabrication remains challenging in laser powder bed fusion (LPBF), while thermomechanical simulation… | ↗Not briefed |
bluesky bluesky-ai | 22h | Remember, kids, you can’t spell ABOMINATION without AI, no matter what Claude tells you | Remember, kids, you can’t spell ABOMINATION without AI, no matter what Claude tells you | ↗Not briefed |
hn hn-ai | 22h | Show HN: Read a Claude Code or Codex session file as a page, in the browser | — | ↗Not briefed |
hn hn-ai | 22h | Exhibit 99.1: Anthropic's Delaware § 3556 Trust and $140B Standing Gate | — | ↗Not briefed |
bluesky bluesky-ai | 22h | "Amazon is betting that many Ring customers will also want an AI-powered security drone and robot patrolling their homes or offices." Who … | "Amazon is betting that many Ring customers will also want an AI-powered security drone and robot patrolling their homes or offices."… | ↗Not briefed |