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 | 22d | experts or anyone interested can u check this out my and open for ur feedbacks MESE-CR: An audit of activation reading and steering for mo… | experts or anyone interested can u check this out my and open for ur feedbacks MESE-CR: An audit of activation reading and steering f… | ↗Not briefed |
bluesky bluesky-ai | 22d | I've played with this sort of stuff a little, and you can get really good sound by having an LLM generate a score then passing it to musehu… | I've played with this sort of stuff a little, and you can get really good sound by having an LLM generate a score then passing it to m… | ↗Not briefed |
bluesky bluesky-ai | 22d | I think the LLM is manipulating MIDI, essentially, and then pulling from something like a sample library for the tones. Or that key presses… | I think the LLM is manipulating MIDI, essentially, and then pulling from something like a sample library for the tones. Or that key pr… | ↗Not briefed |
arxiv arxiv-ai | 22d | Local Evidence, Global Failure: Diagnosing Speech-Vision Temporal Composition in Audio-Visual Language Models🔧 Modal | Audio-visual large language models (AV-LLMs) can fail at full-video speech-vision decisions even when useful visual evidence is locall… | ↗Not briefed |
bluesky bluesky-ai | 22d | "okay with the proper usage of LLM-" no *stabs* | "okay with the proper usage of LLM-" no *stabs* | ↗Not briefed |
bluesky bluesky-ai | 22d | Anthropic reports that junior technical hires have fallen from 3 in 10 to 1 in 10 as AI adoption changes its workforce. This reflects an in… | Anthropic reports that junior technical hires have fallen from 3 in 10 to 1 in 10 as AI adoption changes its workforce. This reflects… | ↗Not briefed |
arxiv arxiv-ai | 22d | Predicted Class Correction: Conformal Thresholds for Classes with No Calibration Data | Class-conditional conformal prediction calibrates a separate score threshold for every class and therefore requires labelled calibrati… | ↗Not briefed |
hn hn-ai | 22d | Ask HN: System One models for autonomous agents | I've seen a few impressive demos of Typesafe's new paradigm with System One models, and it seemed really clear to me that th… | ↗Not briefed |
hn hn-ai | 22d | Simultaneous high-fidelity single-qubit gates in a spin qubit array | — | ↗Not briefed |
bluesky bluesky-ai | 22d | As someone who has never knowingly chosen to use AI (altho it gets pushed at us all the time), never used a chatbot, ChatGPT, Claude, or an…🔧 ChatGPT | As someone who has never knowingly chosen to use AI (altho it gets pushed at us all the time), never used a chatbot, ChatGPT, Claude,… | ↗Not briefed |
bluesky bluesky-ai | 22d | DeepSeek 4.1-flash-next came out yesterday, and in some cases it matches or surpases the best frontier models from just a few months ago. …🔧 GPT-5🔧 Qwen 3 | DeepSeek 4.1-flash-next came out yesterday, and in some cases it matches or surpases the best frontier models from just a few months a… | ↗Not briefed |
arxiv arxiv-ai | 22d | Formally Verified Autoresearch for Analyzing the Expressivity of Multi-Head Attention | Multi-head attention is a central component of the transformer architecture, yet how its expressivity depends on the number of heads r… | ↗Not briefed |
bluesky bluesky-ai | 22d | 🤔 thenextweb.com/news/daniela... | 🤔 thenextweb.com/news/daniela... | ↗Not briefed |
rss techmeme | 22d | Sources: HomeView units sat in warehouses for months; Apple plans a Nov. 6 store reset, likely for new Macs; iPhone Duo preorder mess could boost 18 Pro sales (Mark Gurman/Bloomberg) | Mark Gurman / Bloomberg: Sources: HomeView units sat in warehouses for months; Apple plans a Nov. 6 store reset, likely for new Macs;… | ↗Not briefed |
arxiv arxiv-ai | 22d | A Distribution-Free Separation Between Learned and Fixed Features | The success of neural networks is largely attributed to their ability to learn useful representations from training data. However, kno… | ↗Not briefed |
bluesky bluesky-ai | 22d | E9 Så sant eg kan unngå det, skal eg unngå generativ KI og LLM og dets like. Eg vil ha maskiner til å gjera drittarbeidet mitt. Ta oppvas… | E9 Så sant eg kan unngå det, skal eg unngå generativ KI og LLM og dets like. Eg vil ha maskiner til å gjera drittarbeidet mitt. Ta o… | ↗Not briefed |
hn hn-ai | 22d | Show HN: Secret Guard – a local LLM gateway; keys never leave your machine | — | ↗Not briefed |
hn hn-ai | 22d | Show HN: Gecit, share a Windows PC's keyboard and mouse with a Mac over QUIC | — | ↗Not briefed |
bluesky bluesky-ai | 22d | tired: vibecoding wired: vibecompiling introducing neurastheniacc, an LLM-based compiler from C to WASM! finally we can be free of our lon… | tired: vibecoding wired: vibecompiling introducing neurastheniacc, an LLM-based compiler from C to WASM! finally we can be free of ou… | ↗Not briefed |
arxiv arxiv-ai | 22d | Follow the Mixed Groups: Ridge-Tracking Gaussian Sampling for Zero-Variance Web Rollouts | Group relative policy optimization learns from a question only when the sampled responses disagree. A group that is entirely correct o… | ↗Not briefed |
hn hn-ai | 22d | Show HN: BricksChat – Generate Lego builds from images, including Technic🔧 Modal | I've been working on generative LEGO assembly models recently, but there are similar projects already, so I wanted to try some ha… | ↗Not briefed |
bluesky bluesky-ai | 22d | Every single one of these companies is a defense contractor. And I can say very confidently that at least Microsoft is actively working wit… | Every single one of these companies is a defense contractor. And I can say very confidently that at least Microsoft is actively workin… | ↗Not briefed |
hn hn-ai | 22d | Omarchy Gets the Web Browser Launching Faster by Avoiding XDG Overhead | — | ↗Not briefed |
arxiv arxiv-ai | 22d | Where Groups Split: A Probe Prior for Nonzero Within-Group Variance in RL Post-Training | Group relative policy optimization learns from a prompt only when the rewards in the sampled group disagree. With a binary verifier, a… | ↗Not briefed |
bluesky bluesky-ai | 22d | An ad from 2054, found early. In 2025 Mattel and OpenAI announced A.I. toys. What would your kids ask it first? | An ad from 2054, found early. In 2025 Mattel and OpenAI announced A.I. toys. What would your kids ask it first? | ↗Not briefed |
bluesky bluesky-ai | 22d | I think a lot about how we develop models of the people we know, get really good at predicting how they'll react, and how it's like having … | I think a lot about how we develop models of the people we know, get really good at predicting how they'll react, and how it's like ha… | ↗Not briefed |
arxiv arxiv-ai | 22d | Where and Why Search-Augmented LLMs Fail When Reasoning over Rapidly Evolving Knowledge | As the world changes, new information is published while older records remain online. This complicates reasoning for large language mo… | ↗Not briefed |
hn hn-ai | 22d | GhostOS – Lightweight, Offline Desktop AI Assistant (macOS/Win/Linux) | — | ↗Not briefed |
bluesky bluesky-ai | 22d | occasionally i show an LLM my sleep data and they react like i was bitten by a radioactive narcoleptic and became Sleep-Recovery-Man consi… | occasionally i show an LLM my sleep data and they react like i was bitten by a radioactive narcoleptic and became Sleep-Recovery-Man… | ↗Not briefed |
bluesky bluesky-ai | 22d | 10/19 「海外旅行の日」 「TOEICの日」 ですよ✨ スマホで通訳してもらうもんっ✨ 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス atelier.fan #AIイラスト #ATelie… | 10/19 「海外旅行の日」 「TOEICの日」 ですよ✨ スマホで通訳してもらうもんっ✨ 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス atelier.fan #AIイラスト #A… | ↗Not briefed |
bluesky bluesky-ai | 22d | まだ眠くないけど（夜行性） 明日からまた一週間がんばろっ💞 おやすみ世界👼🪽 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス atelier.fan #AIイラスト #ATelier #A… | まだ眠くないけど（夜行性） 明日からまた一週間がんばろっ💞 おやすみ世界👼🪽 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス atelier.fan #AIイラスト #ATeli… | ↗Not briefed |
bluesky bluesky-ai | 22d | LLMのせいで疑心暗鬼になったら、美術館のサイトに行ってパブリックドメインになっている数百年前の絵や彫刻を見ることにしてる。 ↓日本の国立美術館5つが協働している目録検索システム search.artmuseums.go.jp | LLMのせいで疑心暗鬼になったら、美術館のサイトに行ってパブリックドメインになっている数百年前の絵や彫刻を見ることにしてる。 ↓日本の国立美術館5つが協働している目録検索システム search.artmuseums.go.jp | ↗Not briefed |
bluesky bluesky-ai | 22d | Klagt bitte! www.derstandard.at/story/300000... | Klagt bitte! www.derstandard.at/story/300000... | ↗Not briefed |
arxiv arxiv-ai | 22d | AdaFedXplain: Adaptive Federated Learning with Explainable AI for Ransomware Detection | Attacks using ransomware have developed into complex multiphase campaigns that target vital infrastructure, resulting in annual damage… | ↗Not briefed |
hn hn-ai | 22d | RentAHuman | — | ↗Not briefed |
arxiv arxiv-ai | 22d | Atoms, Not Knowledge: What Production-Scale Addressed-Memory Writes Deposit | Deterministic hash-addressed conditional memory permits local post-hoc knowledge injection. Extending arXiv:2610.21553's bounded-feasi… | ↗Not briefed |
hn hn-ai | 22d | Show HN: ElNeo, a E2E enabled matrix client for Emacs, inspired by telega | About a year ago I was wondering how hard it'd be to build an e2e enabled emacs matrix client similar to how telega does things:… | ↗Not briefed |
bluesky bluesky-ai | 22d | Sam Altman (CEO of OpenAI) thinks "a significant fraction of the power on Earth should be running AI compute." But even the guys building i… | Sam Altman (CEO of OpenAI) thinks "a significant fraction of the power on Earth should be running AI compute." But even the guys build… | ↗Not briefed |
bluesky bluesky-ai | 22d | If you've been following me for the last few months, I've been tweaking a local LLM stack to run on my MacBook Pro. I decided to write up a… | If you've been following me for the last few months, I've been tweaking a local LLM stack to run on my MacBook Pro. I decided to write… | ↗Not briefed |
hn hn-ai | 22d | Nordwand Mono – A monospace font for coders and adventurers | — | ↗Not briefed |
hn hn-ai | 22d | World Bank says 'small AI' could transform developing countries | — | ↗Not briefed |
arxiv arxiv-ai | 22d | Shared-Private Cross-Modal Discrepancy Learning for Imbalanced Multi-Source Audio-Visual Deepfake Detection🔧 Modal | Audio-visual deepfake detectors are often trained under a single dataset or manipulation regime, but multi-source training collections… | ↗Not briefed |
arxiv arxiv-ai | 22d | Learning When to Abort in Transformer Planning: A Mechanistic Study | Planning does not by itself imply the ability to recognise errors, but recognising errors can improve planning capacity. In this paper… | ↗Not briefed |
arxiv arxiv-ai | 22d | Aligning Language Models with Indigenous Community Cosmovision through Constitutional AI | Large language models (LLMs) are predominantly trained on corpora that reflect Western, anglophone worldviews. This asymmetry is not n… | ↗Not briefed |
hn hn-ai | 22d | Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages | — | ↗Not briefed |
arxiv arxiv-ai | 22d | World Model Interpretability: From Decoding Representations to Understanding Latent Worlds | Emerging Joint-Embedding Predictive Architectures (JEPAs) are moving beyond representation learning towards latent world models that s… | ↗Not briefed |
hn hn-ai | 22d | When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers | — | ↗Not briefed |
hn hn-ai | 22d | GhostAction hid a credential stealer in a fake font file in 94 GitHub repos | — | ↗Not briefed |
hn hn-ai | 22d | Show HN: An open-source curation of most useful ChatGPT Plugins🔧 ChatGPT | — | ↗Not briefed |
bluesky bluesky-ai | 22d | Maybe we’ll all get so bad at stuff, we’ll think the LLM is great at everything? | Maybe we’ll all get so bad at stuff, we’ll think the LLM is great at everything? | ↗Not briefed |