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 | 17d | GraphCast-Mamba: Learning Non-Markovian Corrections for Weather Forecasting | Unresolved atmospheric processes can introduce history dependence beyond the short input windows used by many weather forecasting mode… | ↗Not briefed |
rss techmeme | 17d | Sources: Polymarket investors pressured CEO Shayne Coplan to add stricter financial and operational controls; Elon Musk's lawyer, Alex Spiro, joined its board (Bloomberg) | Bloomberg: Sources: Polymarket investors pressured CEO Shayne Coplan to add stricter financial and operational controls; Elon Musk's l… | ↗Not briefed |
bluesky bluesky-ai | 17d | I always get very confused about using (augmented) commercial models like Claude for experiments like this. Surely you need to know everyth… | I always get very confused about using (augmented) commercial models like Claude for experiments like this. Surely you need to know ev… | ↗Not briefed |
bluesky bluesky-ai | 17d | strongly recommend reading this, regardless of whether you’re an AI fan,skeptic, user, whatever. Among other things, this has a diagram of … | strongly recommend reading this, regardless of whether you’re an AI fan,skeptic, user, whatever. Among other things, this has a diagra… | ↗Not briefed |
arxiv arxiv-ai | 17d | Variance Reduction with Delayed Feedback for Non-Convex Optimization | We study variance reduction for non-convex optimization when gradient evaluations return with a delay. The server can update the param… | ↗Not briefed |
arxiv arxiv-ai | 17d | ECG for laboratory values:A comprehensive physiological and technical analysis | Laboratory measurements are essential for clinical decision making but typically require invasive blood sampling. Recent research sugg… | ↗Not briefed |
hn hn-ai | 17d | Show HN: we made every cloud work like one giant computer | — | ↗Not briefed |
bluesky bluesky-ai | 17d | DeepSeek Dials Up the Pressure on OpenAI and Anthropic With Open-Source Inference Tool | DeepSeek Dials Up the Pressure on OpenAI and Anthropic With Open-Source Inference Tool | ↗Not briefed |
hn hn-ai | 17d | Show HN: Not Quite My Tempo – A Whiplash-Style AI Reviewer for GitHub PRs | — | ↗Not briefed |
bluesky bluesky-ai | 17d | zeit für den evergreen | zeit für den evergreen | ↗Not briefed |
bluesky bluesky-ai | 17d | Unity devs: I am looking for someone building a game with NPCs to try Spanda, a private Unity SDK I have been developing. It lets existing… | Unity devs: I am looking for someone building a game with NPCs to try Spanda, a private Unity SDK I have been developing. It lets exi… | ↗Not briefed |
arxiv arxiv-ai | 17d | BitMTP: When Multi-Token Prediction Meets Low-Bit Large Language Models | For efficient deployment of large language models (LLMs), methods of ultra-low-bit backbones reduce weight storage, memory traffic, an… | ↗Not briefed |
arxiv arxiv-ai | 17d | Present but Hard to Read: Probing Latent Reasoning with Minimal Pairs | Probing studies of latent reasoning often read a model's state where the answer begins and take a failed probe as evidence that a valu… | ↗Not briefed |
hn hn-ai | 17d | Running Local Models for Agents – A Mac Mini vs. a 6-Year-Old GPU | — | ↗Not briefed |
rss the-decoder | 17d | OpenAI says GPT-6 is much better with legacy programming languages such as COBOL and Fortran🔧 GPT-5 | OpenAI improved its latest models for low-resource programming languages like COBOL, Fortran, and Visual Basic. On the new MultiLangBe… | ↗Not briefed |
hn hn-ai | 17d | Google open-sources TorchTPU, a native PyTorch back end for TPUs | — | ↗Not briefed |
hn hn-ai | 17d | Beyond g-mean²: complementary metrics for monitorability | — | ↗Not briefed |
arxiv arxiv-ai | 17d | Model- and Layer-Level Closed-Loop Forecasting with SSM-Based Spiking Neural Networks | Deep State-Space Models (Deep SSMs) have shown promising performance and computational benefits in processing of time-series data with… | ↗Not briefed |
bluesky bluesky-ai | 17d | Like *even if* (and it feels like a big if) there is a legitimate use for gen AI, I can’t believe art and publishing is it. We are literall… | Like *even if* (and it feels like a big if) there is a legitimate use for gen AI, I can’t believe art and publishing is it. We are lit… | ↗Not briefed |
hn hn-ai | 17d | The Topology of an LLM Agent Loop with Tool Use | — | ↗Not briefed |
arxiv arxiv-ai | 17d | OptionWorld: A Benchmark for Diagnosing Failures of LLMs as Long-Term Personal Assistants | Long-term assistants must follow unfinished tasks across days as information and circumstances change. Finding information or starting… | ↗Not briefed |
bluesky bluesky-ai | 17d | Good information about climate change in four links: www.linkedin.com/news/story/f... | Good information about climate change in four links: www.linkedin.com/news/story/f... | ↗Not briefed |
bluesky bluesky-ai | 17d | me going to work this morning: 😀🎶🎶🎶 me seeing a go patch release for security: eh, probably not a big deal me opening up the patch rele… | me going to work this morning: 😀🎶🎶🎶 me seeing a go patch release for security: eh, probably not a big deal me opening up the patch… | ↗Not briefed |
hn hn-ai | 17d | Linus Torvalds posts his first AI-assisted patch to the Linux kernel | — | ↗Not briefed |
arxiv arxiv-ai | 17d | Asymmetric Looped Transformers: Fixed Global Representations for an Iterative Decoder | Looped Transformers increase effective depth by repeatedly applying shared layers, offering a way to scale computation without increas… | ↗Not briefed |
hn hn-ai | 17d | Hacking Anthropic For Free Unlimited Claude | — | ↗Not briefed |
bluesky bluesky-ai | 17d | Anthropic and Elon Musk’s SpaceX have committed billions of dollars in stocks to benefit the public—all while dodging accountability. | Anthropic and Elon Musk’s SpaceX have committed billions of dollars in stocks to benefit the public—all while dodging accountability. | ↗Not briefed |
bluesky bluesky-ai | 17d | How perfect? How improbable? Well, if Anthropic is forced to IPO at the lower price of $1.5tn, it'll disappoint investors, reduce the amoun… | How perfect? How improbable? Well, if Anthropic is forced to IPO at the lower price of $1.5tn, it'll disappoint investors, reduce the… | ↗Not briefed |
bluesky bluesky-ai | 17d | This piece distills something I've said repeatedly: that AI "working out" requires too many improbable things to happen, and that when the … | This piece distills something I've said repeatedly: that AI "working out" requires too many improbable things to happen, and that when… | ↗Not briefed |
bluesky bluesky-ai | 17d | Profitability, of course, assumes that the customer for the compute exists, and they have customers. Unfortunately, OpenAI and Anthropic ar… | Profitability, of course, assumes that the customer for the compute exists, and they have customers. Unfortunately, OpenAI and Anthrop… | ↗Not briefed |
bluesky bluesky-ai | 17d | In fairness, the whole "abandoning reason" has seemingly already happened. This late in the bubble, there is no consensus on how much a gig…🔧 Consensus | In fairness, the whole "abandoning reason" has seemingly already happened. This late in the bubble, there is no consensus on how much… | ↗Not briefed |
arxiv arxiv-ai | 17d | Geometry-Agnostic Data-Driven Discovery of Extreme Scenarios | We introduce a two-stage data-driven neural framework for discovering dynamically admissible extreme scenarios. Rather than assuming p… | ↗Not briefed |
bluesky bluesky-ai | 17d | Borderless AI brain and talent drain... www.cbsnews.com/news/ai-doom... #ai #llm #trained #staff vs #AIslop #brainDrain #talent #loss #Pe… | Borderless AI brain and talent drain... www.cbsnews.com/news/ai-doom... #ai #llm #trained #staff vs #AIslop #brainDrain #talent #los… | ↗Not briefed |
bluesky bluesky-ai | 17d | DeepSeek infer is the cloud version of DwarfStar, conceptually. Vertical inference engine for one model and specific hardware. Good move. A… | DeepSeek infer is the cloud version of DwarfStar, conceptually. Vertical inference engine for one model and specific hardware. Good mo… | ↗Not briefed |
bluesky bluesky-ai | 17d | The torture of the interrogation—an attack on the disabled?—and the despair present in her failure to meet the moment, which suggested that… | The torture of the interrogation—an attack on the disabled?—and the despair present in her failure to meet the moment, which suggested… | ↗Not briefed |
arxiv arxiv-ai | 17d | Proof-Carrying Contracts: Commitments AI Agents Cannot Break | AI agents increasingly act on behalf of people or other agents, purchasing services, exchanging information and delegating work. What… | ↗Not briefed |
arxiv arxiv-ai | 17d | Exploring Temporal Context-Aware Physical Grounding in Latent World Models | Learning latent dynamics is fundamental to building action-conditioned world models for planning. Among existing families, Joint Embed… | ↗Not briefed |
bluesky bluesky-ai | 17d | "Slop and vibe-coded games are a scourge, AI is a technofascist, anti-human and anti-art project—" So to be clear, if you start a blog pos… | "Slop and vibe-coded games are a scourge, AI is a technofascist, anti-human and anti-art project—" So to be clear, if you start a blo… | ↗Not briefed |
arxiv arxiv-ai | 17d | Claims of emergence in AI should be grounded in information decomposition | When a language model develops capabilities not traceable to its individual components, when a multi-agent AI system outperforms each… | ↗Not briefed |
hn hn-ai | 17d | Show HN: Git-Zmarshal | Orchestrating build trains for many concurrent agents, on top of git. | ↗Not briefed |
hn hn-ai | 17d | Testing Jev, a fast multiple-choice AI model, on forensic accounting tasks | — | ↗Not briefed |
hn hn-ai | 17d | Ask HN: My Gemini bill shot up suddenly to 2500 USD in one day | our usage of gemini is very limited. using 2.5 flash or similar models. at a hackathon in bay area i used the api key in my google app… | ↗Not briefed |
hn hn-ai | 17d | Best LLM Fine-Tuning Services: What to Look for in 2026 | — | ↗Not briefed |
arxiv arxiv-ai | 17d | Beyond Trajectory Statistics: Community-Consistent Human Mobility Generation via Evolving User-Location Graphs | Human mobility data are fundamental to Web-based location services and broader urban applications, and mobility generation provides sy… | ↗Not briefed |
bluesky bluesky-ai | 17d | Direct link to opt out of eBay genAI training model accountsettings.ebay.co.uk/ai-preferences | Direct link to opt out of eBay genAI training model accountsettings.ebay.co.uk/ai-preferences | ↗Not briefed |
hn hn-ai | 17d | Show HN: armsense – firewall & router ARM64 distro based on pfSense and FreeBSD | — | ↗Not briefed |
hn hn-ai | 17d | Terrifi: A Terraform provider for UniFi (built with Claude and HITL testing) | — | ↗Not briefed |
bluesky bluesky-ai | 17d | For example, Anthropic’s offline monitors would have missed Mythos 5’s supply chain attack during recent cyber incidents, despite Anthropic… | For example, Anthropic’s offline monitors would have missed Mythos 5’s supply chain attack during recent cyber incidents, despite Anth… | ↗Not briefed |
arxiv arxiv-ai | 17d | LocaLeJEPA: Maximum Entropy Representations under Geometric Priors | Joint-Embedding Predictive Architectures (JEPA) offer an appealing approach to self-supervised learning, but are typically prone to re… | ↗Not briefed |
arxiv arxiv-ai | 17d | GPT-6-Astra Understands Touch and Outperform Vision--Tactile--Language--Action Models | General-purpose multimodal models have shown promise as direct robot action policies, yet whether they grasp the physics underlying re… | ↗Not briefed |