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 | 2d | BREAKING: Sam Altman’s OpenAI says it will IPO “sometime." | BREAKING: Sam Altman’s OpenAI says it will IPO “sometime." | ↗Not briefed |
hn hn-ai | 2d | Show HN: MET – a map of the people you've met while traveling | — | ↗Not briefed |
rss simon-willison | 2d | Quoting Daniel Stenberg | — | ↗Not briefed |
bluesky bluesky-ai | 2d | rubbing a hole through my temples reading the guy's justifications on his LLM translation of the snatcher ports. if you spent a modicum of … | rubbing a hole through my temples reading the guy's justifications on his LLM translation of the snatcher ports. if you spent a modicu… | ↗Not briefed |
hn hn-ai | 2d | Show HN: A simple Kanban job application tracker | I stepped away from tech for a few years, and am on my way back in. Wrote this simple application tracker because I like Kanban a lot,… | ↗Not briefed |
bluesky bluesky-ai | 2d | I have just collected "Atheria Deviation" by @javiergalanrico.bsky.social on @objkt.com https://objkt.com/tokens/KT1RJ6PbjHpwc3M5rw5s2Nbmef… | I have just collected "Atheria Deviation" by @javiergalanrico.bsky.social on @objkt.com https://objkt.com/tokens/KT1RJ6PbjHpwc3M5rw5s2… | ↗Not briefed |
arxiv arxiv-ai | 2d | CurvDelta: Low-Curvature Delta Modeling For Surgical Learning and Unlearning | A central challenge in adapting large language models (LLMs) is capability preservation: an adaptation can achieve its target while de… | ↗Not briefed |
bluesky bluesky-ai | 2d | LLM reply bots are annoying because at first glance you're like "this seems like a legit response" but then once you've read enough bot res… | LLM reply bots are annoying because at first glance you're like "this seems like a legit response" but then once you've read enough bo… | ↗Not briefed |
arxiv arxiv-ai | 2d | Piccolo: Co-Evolving Decision Models and Data through Recursive Self-Improvement | We present Piccolo, a family of multimodal decision models (0.8B to 27B) that coevolves model capabilities and training data via Agent… | ↗Not briefed |
arxiv arxiv-ai | 2d | Self-Rewarded Reinforcement Learning for Better-Structured Medical Explanations in Small Language Models | Small language models (SLMs) can run on commodity hardware and support privacy-preserving, on-premises clinical deployment, but their… | ↗Not briefed |
arxiv arxiv-ai | 2d | Zonotopic Analysis of Quantization Error in Neural Networks | Quantized neural networks achieve computational efficiency by reducing precision, but bounding their quantization error remains a chal… | ↗Not briefed |
bluesky bluesky-ai | 2d | Some wonder how Anthropic employees average $72M in equity. Here’s how. Business Insider profiled a midlevel 2024 hire whose $1.3M grant w… | Some wonder how Anthropic employees average $72M in equity. Here’s how. Business Insider profiled a midlevel 2024 hire whose $1.3M gr… | ↗Not briefed |
bluesky bluesky-ai | 2d | OpenAI Board Member Warns Public They Could Be “Dead or Enslaved” in One Year "For most people in this room, the most likely reason they wi… | OpenAI Board Member Warns Public They Could Be “Dead or Enslaved” in One Year "For most people in this room, the most likely reason th… | ↗Not briefed |
bluesky bluesky-ai | 2d | Крутой открытый софт: MiniMind обучает 64M LLM с нуля за 2 часа и 3 бакса. Полный цикл, чистый PyTorch. Разбор в Telegram: https://t.me/the… | Крутой открытый софт: MiniMind обучает 64M LLM с нуля за 2 часа и 3 бакса. Полный цикл, чистый PyTorch. Разбор в Telegram: https://t.m… | ↗Not briefed |
hn hn-ai | 2d | Show HN: I built an infinite canvas to explore 340k artworks from The Met | Hey HN!<p>I built this project, Open Metropolitan, which is a digital museum that uses The Met's Open Access data to give people… | ↗Not briefed |
arxiv arxiv-ai | 2d | Equal Steps, Unequal Curvature: Where Muon's Update Goes Wrong | Muon updates each hidden weight matrix along the polar factor of its momentum, the steepest-descent direction under the spectral norm.… | ↗Not briefed |
arxiv arxiv-ai | 2d | Choosing the Right Representations: Task-Aware Fusion for Multimodal Materials Language Models | Multimodal language models for materials science rely on effective mechanisms for incorporating three-dimensional structural represent… | ↗Not briefed |
bluesky bluesky-ai | 2d | Every time we play a session I think about this one person I worked with who told me his friend's text adventure indie game involved an LLM… | Every time we play a session I think about this one person I worked with who told me his friend's text adventure indie game involved a… | ↗Not briefed |
rss the-decoder | 2d | Anthropic, OpenAI, and Google models readily marked medics, journalists, and surrendering soldiers as kamikaze drone targets | A Vals AI investigation shows that leading AI models from Anthropic, OpenAI and Google have almost no safeguards against use in autono… | ↗Not briefed |
bluesky bluesky-ai | 2d | In der #heiseshow: Anthropic plant den Börsengang, die Abhängigkeit von Microsoft, Apple & Co. wird zum Risiko und Voyager 1 ist einen Lich… | In der #heiseshow: Anthropic plant den Börsengang, die Abhängigkeit von Microsoft, Apple & Co. wird zum Risiko und Voyager 1 ist einen… | ↗Not briefed |
arxiv arxiv-ai | 2d | Hypnos: A Dual-Phase End-to-End Platform for Scalable Continual Learning at the Edge | Continual learning (CL) enables neural networks to adapt to non-stationary data streams by mitigating catastrophic forgetting, offerin… | ↗Not briefed |
bluesky bluesky-ai | 2d | And while we’re on the topic, fanfiction/fanart is WORK. It’s not “content” to be consumed. It is WORK. It is fan WORK. I'm not opening an … | And while we’re on the topic, fanfiction/fanart is WORK. It’s not “content” to be consumed. It is WORK. It is fan WORK. I'm not openin… | ↗Not briefed |
arxiv arxiv-ai | 2d | Clover-35B-A3B-Preview Technical Report | We introduce Clover-35B-A3B-Preview (Clover), a mixture-of-experts diffusion language model with 35 billion total parameters and 3 bil… | ↗Not briefed |
arxiv arxiv-ai | 2d | H3Remover: Reinforcing Video Object Removal without Paired Data | Pretrained video editors possess rich removal and completion priors, yet may retain targets or alter unrelated content. We present H3R… | ↗Not briefed |
bluesky bluesky-ai | 2d | I have just collected "Lacuna" by @vailamy.objkt.social on @objkt.com https://objkt.com/tokens/KT1T3KFD4tHX5Yw6VsnPnyx56z1pS8LGVMg8/7?ref=t… | I have just collected "Lacuna" by @vailamy.objkt.social on @objkt.com https://objkt.com/tokens/KT1T3KFD4tHX5Yw6VsnPnyx56z1pS8LGVMg8/7?… | ↗Not briefed |
hn hn-ai | 2d | AMD OpenSIL Achieves Its Production Goal with 6th Gen EPYC "Venice" | — | ↗Not briefed |
rss theverge-ai | 2d | Remember the frontier AI pause? Chip companies don’t either🔧 Hugging Face | In October, Nvidia began shipping its most advanced AI computing system, the Vera Rubin NVL72, calling it the “fastest product ramp in… | ↗Not briefed |
bluesky bluesky-ai | 2d | Morgen all in Anthropic und danach reich oder Privatinsolvenz. | Morgen all in Anthropic und danach reich oder Privatinsolvenz. | ↗Not briefed |
bluesky bluesky-ai | 2d | Anthropic goes public this week. But is the AI company part of a massive bubble just waiting to pop? Hear tech journalist @byjacobward.bs… | Anthropic goes public this week. But is the AI company part of a massive bubble just waiting to pop? Hear tech journalist @byjacobwa… | ↗Not briefed |
hn hn-ai | 2d | Show HN: Wifehash; a 128-bit hash that beats XXH3-128 on bulk data | - XXH3-128 is still faster on short keys - Competes with PolyXOR128 on Alder Lake - Doesn't use crypto extensions (AESNI) or carr… | ↗Not briefed |
hn hn-ai | 2d | Recent Advances in the Use of LLMs to Solve Mixed-Integer Programming Models | — | ↗Not briefed |
hn hn-ai | 2d | I'm worried about LLMs accelerating hardware for cheap ASI | — | ↗Not briefed |
bluesky bluesky-ai | 2d | If the Anthropic IPO goes to the moon I am deleting my Blue Sky account. And will never read CNBC news again. Wall Street | If the Anthropic IPO goes to the moon I am deleting my Blue Sky account. And will never read CNBC news again. Wall Street | ↗Not briefed |
arxiv arxiv-ai | 2d | Transferring an analysis of human brain organization to language models | During human cortical development, functional systems become increasingly differentiated. We asked whether improved reasoning in langu… | ↗Not briefed |
hn hn-ai | 2d | AI is writing a public web every 15 days | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: Break a robot arm and watch a meta-RL agent keep writing | — | ↗Not briefed |
arxiv arxiv-ai | 2d | Evaluating the Effectiveness of Location Encoding Methods for Species Distribution Modelling | Understanding the spatial distribution of species is a key question in ecology. A variety of solutions, from statistical, traditional… | ↗Not briefed |
arxiv arxiv-ai | 2d | GateWatch: Low-Overhead LLM Monitoring with Boolean Circuits🔧 vLLM | As large language models (LLMs) and LLM-based agents are increasingly deployed, monitoring their behavior becomes important for detect… | ↗Not briefed |
hn hn-ai | 2d | Open-source mock MCP server to test clients and agents without real servers | — | ↗Not briefed |
hn hn-ai | 2d | Anthropic's amended S-1 in 19 slides | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: AgentKombat – LLM agents fight; your prompt is the fighting style | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: MIT's How to Make (Almost) Anything as a 100-day email course | A recent HN post linked to MIT’s famous How To Make (Almost) Anything course materials which I have wished I could have taken back in… | ↗Not briefed |
arxiv arxiv-ai | 2d | Negotiating Control and Accountability among Stakeholders in High-Risk AI Use Cases: Development and Evaluation of the Method ALIGN | Characteristics of Artificial Intelligence (AI) severely restrict human control. This creates new challenges for socio-technical syste… | ↗Not briefed |
bluesky bluesky-ai | 2d | "erm... the author doesn't owe You anything! 🤓☝️" yes they do, anthropic now owes me $100 for reading that | "erm... the author doesn't owe You anything! 🤓☝️" yes they do, anthropic now owes me $100 for reading that | ↗Not briefed |
bluesky bluesky-ai | 2d | Introducing Raycast MCP 🔌 Use Raycast from any agent: 📂 Access your Notes, Clipboard History and more ⌨️ Automate your setup, like assig… | Introducing Raycast MCP 🔌 Use Raycast from any agent: 📂 Access your Notes, Clipboard History and more ⌨️ Automate your setup, like… | ↗Not briefed |
rss techmeme | 2d | An interview with philosopher Thomas Nagel, who won a $1M prize for contributions to consciousness and says the notion of AI consciousness is "just ridiculous" (Steven Levy/Wired) | Steven Levy / Wired: An interview with philosopher Thomas Nagel, who won a $1M prize for contributions to consciousness and says the n… | ↗Not briefed |
arxiv arxiv-ai | 2d | Diffusion Models are Provably Efficient Priors for Sparse Recovery | Diffusion models have transformed generative modeling through their ability to capture and generate samples from complex data distribu… | ↗Not briefed |
hn hn-ai | 2d | Show HN: FLAC, MP3 and DST decoders, about 2x faster than FFmpeg's (MIT) | I've published three audio decoders under the MIT licence: FLAC, MP3, and DST (the lossless compression inside SACDs). C++17, no… | ↗Not briefed |
bluesky bluesky-ai | 2d | Heck, even an LLM could parse the fundamental error in that headline. | Heck, even an LLM could parse the fundamental error in that headline. | ↗Not briefed |
bluesky bluesky-ai | 2d | Microsoft, NSA, bancos e ataques hackers: veja a cronologia dos apagões digitais entre 12 e 17 de novembro: Uma série de interrupções tecno… | Microsoft, NSA, bancos e ataques hackers: veja a cronologia dos apagões digitais entre 12 e 17 de novembro: Uma série de interrupções… | ↗Not briefed |