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 |
|---|---|---|---|---|
hn hn-ai | 28d | Show HN: Diaktoros | — | ↗Not briefed |
hn hn-ai | 28d | Show HN: An Email-to-iMessage Gateway | Mobile carriers have long supported email-to-SMS functionality, but in the United States, the big three (Verizon, T-Mobile, and AT&… | ↗Not briefed |
arxiv arxiv-ai | 28d | Harness Policy Distillation: Aligning Agent Policies with Runtime Semantics | Deploying a general-purpose model as an agent means running it inside a product-specific runtime harness that decides which actions ar… | ↗Not briefed |
bluesky bluesky-ai | 28d | Surely summary is a skill that tests if you understand the text. If you only ever read summaries all you get is someone else’s understandin… | Surely summary is a skill that tests if you understand the text. If you only ever read summaries all you get is someone else’s underst… | ↗Not briefed |
arxiv arxiv-ai | 28d | Long-Horizon RL Needs Muon | Long-horizon agents must coordinate reasoning, tool use, and environment feedback over extended interactions. Learning from sparse out… | ↗Not briefed |
bluesky bluesky-ai | 28d | The SoA: We're going to do the stupidest, worst thing in publishing today. Faber: Get an LLM to hold my cowardice beer. | The SoA: We're going to do the stupidest, worst thing in publishing today. Faber: Get an LLM to hold my cowardice beer. | ↗Not briefed |
arxiv arxiv-ai | 28d | Repairing Speech Access After Text Updates with Frozen Affine Fields | Speech-capable language models inherit most of their knowledge from text-rich language backbones, and continued model maintenance is l… | ↗Not briefed |
rss the-decoder | 28d | Nvidia's Jensen Huang pledges $2 billion for Europe, $250 million in compute for UK safety labs | Nvidia CEO Jensen Huang announced $2 billion in European investments and new stakes in French startups like Doctolib and Entalpic. The… | ↗Not briefed |
hn hn-ai | 28d | Show HN: I built a phone app to gamify fitness with friends | I'm trying to force myself to do more exercise, but I always lose the motivation to do it. So a few friends and I are building a… | ↗Not briefed |
arxiv arxiv-ai | 28d | Adversarial Reinforcement Learning for Robust Planetary Landing | Deep reinforcement learning (DRL) policies for planetary landing and other aerospace guidance, navigation, and control tasks are commo… | ↗Not briefed |
hn hn-ai | 28d | Tell HN: Wise holds my company's $22k for 27 days, then goes silent | Wise closed the business accounts of my Delaware LLC and my Polish sole proprietorship. I'm not fighting the closure, I just want… | ↗Not briefed |
arxiv arxiv-ai | 28d | Contrastive uncertainty learning for robust multi-structure segmentation in lower-pelvic MR | Boundary delineation of anatomical structures in prostate MRI is often compromised by regional ambiguity. Automated segmentation metho… | ↗Not briefed |
bluesky bluesky-ai | 28d | Anthropic warns small AI models could self-replicate through IoT devices arstechnica.com/security/202...🔧 Replicate | Anthropic warns small AI models could self-replicate through IoT devices arstechnica.com/security/202... | ↗Not briefed |
hn hn-ai | 28d | Show HN: An Algol 60 compiler with a LuaJIT runtime | — | ↗Not briefed |
bluesky bluesky-ai | 28d | Even many lowly workers don't seem to understand. I see many software devs embracing AI because they don't want to be left behind. My count… | Even many lowly workers don't seem to understand. I see many software devs embracing AI because they don't want to be left behind. My… | ↗Not briefed |
arxiv arxiv-ai | 28d | Scaling Multilingual Mixture-of-Experts: How Language Coverage Changes Capacity and Compute Allocation | Multilingual models share capacity across languages, so adding languages can hurt performance on existing ones. Mixture-of-experts mod… | ↗Not briefed |
arxiv arxiv-ai | 28d | P2Patch: Exploit-Guided Vulnerability Repair | Large language models are increasingly deployed as autonomous agents for automated program repair, patching software vulnerabilities t… | ↗Not briefed |
arxiv arxiv-ai | 28d | Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection | Anomaly detection is a fundamental task for time-series analytics, with important implications for the downstream performance of many… | ↗Not briefed |
hn hn-ai | 28d | Asterinas 0.19.0 | — | ↗Not briefed |
hn hn-ai | 28d | Show HN: MLX Direct Decider, Mac local decider without fine-tuning | Hi HN, I built this open source project to enable the existing MLX model on your Mac to be used as a decider, without additional fine-… | ↗Not briefed |
arxiv arxiv-ai | 28d | PAWA: Panoramic Action-conditioned World Modeling for Multiple Agents | A multiplayer world model must predict how the independent actions of several players change a shared environment, including changes o… | ↗Not briefed |
arxiv arxiv-ai | 28d | MAEM: Mitigating Recency Bias via a Replay-Free, Capacity-Controlled Dynamic Mixture of Experts for Class-Incremental Graph Learning | Graph Neural Networks (GNNs) excel on graph-structured data but suffer catastrophic forgetting when classes arrive sequentially. In cl… | ↗Not briefed |
hn hn-ai | 28d | Flashcards in ChatGPT🔧 ChatGPT | — | ↗Not briefed |
hn hn-ai | 28d | Bitwise-identical training and inference across GPU vendors – Nvidia, AMD, Apple | — | ↗Not briefed |
bluesky bluesky-ai | 28d | I lost hope that the Shogun 2 remaster would be good as more news filtered out before release, but I underestimated just how much of a cata… | I lost hope that the Shogun 2 remaster would be good as more news filtered out before release, but I underestimated just how much of a… | ↗Not briefed |
hn hn-ai | 28d | Show HN: BScaff – A CLI tool for project scaffolding and snippets | I built BScaff, a CLI tool for saving and reusing project structures, templates, files, and text snippets. It can serve as a URL bookm… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Merlin, Run Your Claude Code and Codex from Slack, Linear and GitHub | Hi HN, I built Merlin to let you use your own Claude Code and Codex subscription from Slack, Linear, and GitHub.<p>There is a desktop… | ↗Not briefed |
bluesky bluesky-ai | 28d | Napster helped people steal IP from corporations. Anthropic/OpenAi/etc helped corporations steal IP from people. | Napster helped people steal IP from corporations. Anthropic/OpenAi/etc helped corporations steal IP from people. | ↗Not briefed |
arxiv arxiv-ai | 28d | The minimal width of neural networks and immersions of real projective space | We investigate the minimal width of deep neural networks that can approximate all continuous functions from $\mathbb{R}^n$ to $\mathbb… | ↗Not briefed |
rss techmeme | 28d | Salience Labs, which develops specialized all-optical circuit switching tech for AI data centers, raised a $125M Series B led by Bullhound Capital (Mike Wheatley/SiliconANGLE) | Mike Wheatley / SiliconANGLE: Salience Labs, which develops specialized all-optical circuit switching tech for AI data centers, raised… | ↗Not briefed |
hn hn-ai | 28d | GitHub Actions has added lock files | — | ↗Not briefed |
bluesky bluesky-ai | 28d | LLM text generation is just the beginning though. The next wave of AI pushed by big tech is "agentic" AI. The idea is not merely that the l… | LLM text generation is just the beginning though. The next wave of AI pushed by big tech is "agentic" AI. The idea is not merely that… | ↗Not briefed |
bluesky bluesky-ai | 28d | These are some thoughts that have been percolating for a while, but @samfr.bsky.social's excellent write up was the push factor for me to f… | These are some thoughts that have been percolating for a while, but @samfr.bsky.social's excellent write up was the push factor for me… | ↗Not briefed |
arxiv arxiv-ai | 28d | Design Space Exploration of Reinforcement Learning Configurations for Post-Legalization Placement | Graph Neural Networks (GNNs) and Reinforcement Learning (RL) have recently been explored for VLSI placement optimization, but their ef… | ↗Not briefed |
hn hn-ai | 28d | Departure Learning Reviews | — | ↗Not briefed |
bluesky bluesky-ai | 28d | I'm going to follow on from @samfr.bsky.social's latest post... Thoughts on AI & the state: why LLM slop is just the beginning, and why *a… | I'm going to follow on from @samfr.bsky.social's latest post... Thoughts on AI & the state: why LLM slop is just the beginning, and w… | ↗Not briefed |
rss techmeme | 28d | Sources: the EU is negotiating with China to lift sanctions on an unnamed Chinese company in exchange for China lifting sanctions on a German ASML supplier (Bloomberg) | Bloomberg: Sources: the EU is negotiating with China to lift sanctions on an unnamed Chinese company in exchange for China lifting san… | ↗Not briefed |
bluesky bluesky-ai | 28d | My pet theory is actually a generalization of that: if there's no ram/gpu available, then you don't have personal computing. You have to us… | My pet theory is actually a generalization of that: if there's no ram/gpu available, then you don't have personal computing. You have… | ↗Not briefed |
hn hn-ai | 28d | Torvalds "unwilling to work without an LLM" at Kernel Maintainers Summit | — | ↗Not briefed |
bluesky bluesky-ai | 28d | Martlet LLM Raketen werden auch im taktischen Flugabwehrsystem „Rapidranger“ verwendet. Dieses Flugabwehrsystem hatte massgeblichen Anteil … | Martlet LLM Raketen werden auch im taktischen Flugabwehrsystem „Rapidranger“ verwendet. Dieses Flugabwehrsystem hatte massgeblichen An… | ↗Not briefed |
hn hn-ai | 28d | Claude can write executables in pure bytes, no language or compiler in between | — | ↗Not briefed |
bluesky bluesky-ai | 28d | 1976年に「コンピューターを使った数学定理の証明」が発表された際に数学者らはどう反応したのか？ https://gigazine.net/news/20261014-1st-time-machines-mathematical-proof/ | 1976年に「コンピューターを使った数学定理の証明」が発表された際に数学者らはどう反応したのか？ https://gigazine.net/news/20261014-1st-time-machines-mathematical-proof/ | ↗Not briefed |
rss techmeme | 28d | AI infrastructure startup Modal raised a $750M Series D led by Accel at a $15.75B post-money valuation, taking its total funding to $1.2B, and passes $1B ARR (Chris Metinko/Axios)🔧 Modal | Chris Metinko / Axios: AI infrastructure startup Modal raised a $750M Series D led by Accel at a $15.75B post-money valuation, taking… | ↗Not briefed |
bluesky bluesky-ai | 28d | He linked Saudi Arabia to Silicon Valley. Then OpenAI found a bug wrapped in toilet paper ft.trib.al/sQ1in8h | He linked Saudi Arabia to Silicon Valley. Then OpenAI found a bug wrapped in toilet paper ft.trib.al/sQ1in8h | ↗Not briefed |
hn hn-ai | 28d | Can LLMs audit a supermarket shelf yet? | — | ↗Not briefed |
bluesky bluesky-ai | 28d | There's no such thing as ethical LLM/AI use in creative fields. The entire thing is built on stolen work. | There's no such thing as ethical LLM/AI use in creative fields. The entire thing is built on stolen work. | ↗Not briefed |
arxiv arxiv-ai | 28d | Scaling Multilingual Safety and Reasoning Evaluation through Reasoning-Guided Translation | Large language models (LLMs) are increasingly deployed across languages, yet their safety and reasoning capabilities remain predominan… | ↗Not briefed |
arxiv arxiv-ai | 28d | Mixed Poisson diffusion models and application to posterior sampling | Diffusion models are a powerful framework for learning complex probability distributions. While widely used diffusion models for conti… | ↗Not briefed |
bluesky bluesky-ai | 28d | This SoA announcement coming at the same time I keep googling "Anthropic payment when" because $1500 is quite a lot of money to me right no… | This SoA announcement coming at the same time I keep googling "Anthropic payment when" because $1500 is quite a lot of money to me rig… | ↗Not briefed |
arxiv arxiv-ai | 28d | Continual pretraining Prostate MRI ViT Encoder for task alignment | Building foundation models for MRI has emerged as a promising direction for advancing medical image analysis and improving accessibili… | ↗Not briefed |