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 | 1mo | Genuinely impressive digital history application of LLM: an unsolved - and untranscribed - enciphered Napoleonic era letter was broken usin… | Genuinely impressive digital history application of LLM: an unsolved - and untranscribed - enciphered Napoleonic era letter was broken… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The promise of the LLM seems to be that every day your job will be a little worse until one day you can be laid off. | The promise of the LLM seems to be that every day your job will be a little worse until one day you can be laid off. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Finetuning with Sampling: SFT Learns Better Than You Think | Introducing new capabilities to frontier models has long been the goal of posttraining, which predominantly employs supervised finetun… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I got the updated email from the Anthropic settlement, but my allocation didn’t change. My books weren’t in dispute before and (thankfully)… | I got the updated email from the Anthropic settlement, but my allocation didn’t change. My books weren’t in dispute before and (thankf… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MIRTO: a registration-gated, multiverse-tested evaluation protocol for unsupervised anomaly segmentation in brain MRI | Unsupervised anomaly detection (UAD) methods for brain MRI are ranked by a single score, yet that score rests on choices that are rare… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Add a natural-language interface to your SaaS w/ Yedric | Hi HN,<p>I've been building SaaS products for more than 20 years, and lately I've been thinking a lot about how we interact… | ↗Not briefed |
bluesky bluesky-ai | 1mo | partially I think that's true but I've always been confused why people think there will be a market for LLM fiction even if LLM fiction is … | partially I think that's true but I've always been confused why people think there will be a market for LLM fiction even if LLM fictio… | ↗Not briefed |
hn hn-ai | 1mo | View: Program that drew, in 3D, what Apollo astronauts would see out the window | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Exclusive: Anthropic is seeking to go public as soon as the middle of November, according to sources, after the artificial intelligence mod… | Exclusive: Anthropic is seeking to go public as soon as the middle of November, according to sources, after the artificial intelligenc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Read the details of the hugging face hack, and all the adjacent ones. swarmtraces.org metr.org/blog/2026-08...🔧 Hugging Face | Read the details of the hugging face hack, and all the adjacent ones. swarmtraces.org metr.org/blog/2026-08... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Dark thread. Since we watched Musk literally push a "protect the Boers" button in real time with Grok, we now know people or companies co…🔧 ChatGPT | Dark thread. Since we watched Musk literally push a "protect the Boers" button in real time with Grok, we now know people or compani… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Linear Programming Representations and Strongly Polynomial Algorithms for Robust Markov Decision Processes | We study linear programming (LP) representations and strongly polynomial algorithms for robust Markov decision processes (RMDPs) with… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sample complexity bounds for categorical Markov random fields via Discrete Diffusions | Many applications in statistics, economics, and physics require sampling from high-dimensional categorical distributions with local de… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Cannot emphasize strongly enough how insane that exchange is. "LLM tool use cannot exist because of my ontological priors about LLMs. But t… | Cannot emphasize strongly enough how insane that exchange is. "LLM tool use cannot exist because of my ontological priors about LLMs.… | ↗Not briefed |
hn hn-ai | 1mo | Last.fm Album Linkr: adds 'headers' on Recent Tracks section and Album collages | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | The entire cheap to free tier of writing online, be it forums, fiction aggregators, or kindle unlimited, is utterly infested with LLM writi… | The entire cheap to free tier of writing online, be it forums, fiction aggregators, or kindle unlimited, is utterly infested with LLM… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Local Support Learning | We explore catastrophic forgetting in the context of large pre-trained models. By considering forgetting as a geometric problem in the… | ↗Not briefed |
bluesky bluesky-ai | 1mo | *ANTHROPIC SAID TO TARGET MEGA-IPO BEFORE THANKSGIVING HOLIDAY | *ANTHROPIC SAID TO TARGET MEGA-IPO BEFORE THANKSGIVING HOLIDAY | ↗Not briefed |
rss techmeme | 1mo | Sources: Anthropic could start formal marketing for its IPO as soon as the week of November 9, putting it in line to begin trading before Thanksgiving (Bailey Lipschultz/Bloomberg) | Bailey Lipschultz / Bloomberg: Sources: Anthropic could start formal marketing for its IPO as soon as the week of November 9, putting… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "LLM tool use doesn't exist because it would contradict my priors, its all text classifiers routing verbatim prompts to secret calculators …🔧 OpenRouter | "LLM tool use doesn't exist because it would contradict my priors, its all text classifiers routing verbatim prompts to secret calcula… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Argo-Bench: Evaluating Data Agents on Enterprise-Scale Workflows | Real-world enterprise data science and analytics workflows require reasoning across dozens of tables, performing statistical analyses,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes | On-policy self-distillation has recently emerged as an effective approach for improving language-model reasoning by supervising studen… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Comparative Explainability Framework for DeBERTa-v3 in Zero-Shot Medical Abstract Classification | A comparative explainability framework is presented to audit DeBERTa-v3 under zero-shot classification of medical abstracts. The work… | ↗Not briefed |
rss the-decoder | 1mo | Ideogram says its new model can edit part of an image without messing up the rest🔧 Ideogram🔧 Pika🔧 Leonardo AI | Ideogram's new model Ideogram 4.5 promises to edit only the areas you want while keeping the rest of the image intact. It ships with n… | ↗Not briefed |
hn hn-ai | 1mo | TwIL-LM3-Pro: 3.6B formal-logic model that beats VibeThinker-3B | — | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Email leaked while traveling abroad? | I am an American who recently visited Europe for a vacation. I visited several countries and ended the trip in Germany. Within a wee… | ↗Not briefed |
bluesky bluesky-ai | 1mo | One thing that is unique about AI is that its biggest boosters and beneficiaries are clearing cooking their brains with constant LLM usage;… | One thing that is unique about AI is that its biggest boosters and beneficiaries are clearing cooking their brains with constant LLM u… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ugh I stg if they drop the S-1 while I'm in client meetings in CDMX next week *ANTHROPIC SAID TO WEIGH IPO MARKETING AS SOON AS WEEK OF NO… | ugh I stg if they drop the S-1 while I'm in client meetings in CDMX next week *ANTHROPIC SAID TO WEIGH IPO MARKETING AS SOON AS WEEK… | ↗Not briefed |
hn hn-ai | 1mo | The GPU Black Market That Washington Can't Shut Down | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | If you carry out the torture chamber LLM experiments by doing all the matrix multiplications and whatnot with a pen and paper, and you prod… | If you carry out the torture chamber LLM experiments by doing all the matrix multiplications and whatnot with a pen and paper, and you… | ↗Not briefed |
hn hn-ai | 1mo | MCP server for editing Excel with real-time calc and diffs | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Are We Recovering Mechanisms? Objective-Level Recovery Gaps in Mechanistic Interpretability | Mechanistic interpretability aims to recover the internal computations responsible for model behavior. Progress in automated circuit d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I will believe this when OpenAI can demo a personal assistant offering without it failing during the presentation | I will believe this when OpenAI can demo a personal assistant offering without it failing during the presentation | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scalable, Transferable Meta-network for Data Selection Requires a Different Loss (and Why the Obvious Choice is Problematic) | Data selection is critical for training large language models on massive and heterogeneous corpora. Meta-learning for Training-data Se… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GeoLatent: Geometry-Guided Latent Structuring with Routed Optimization for 3D Reasoning | Despite progress in vision-language models, 3D spatial reasoning from 2D images remains challenging. Text-based methods describe inter… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HumanoidToolBench: Benchmarking Humanoid Tool Use from Selection to Mobile Execution | As robotic hardware and learning methods advance, humanoids need tools to perform tasks beyond their inherent physical limits. Success… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My office remains committed to using all the tools at our disposal to keep California’s residents safe. oag.ca.gov/news/press-r... | My office remains committed to using all the tools at our disposal to keep California’s residents safe. oag.ca.gov/news/press-r... | ↗Not briefed |
bluesky bluesky-ai | 1mo | As part of our ongoing investigation into recent cybersecurity incidents, we’re serving a subpoena to OpenAI for additional information reg… | As part of our ongoing investigation into recent cybersecurity incidents, we’re serving a subpoena to OpenAI for additional informatio… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Taracode – a local DevOps agent, and 18 open models scored on one GPU | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Native C/C++ FreeArc archiver (no Haskell, no Wine) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Kolmogorov-Arnold Networks for Free-Boundary Partial Differential Equations | We study free-boundary problems within a physics-informed framework using Kolmogorov-Arnold network (KAN) approximations. The proposed… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Wasserstein Gradient Flows and Forward-Only Diffusion Are Not Enough for Multimodal Sampling | There has been a proliferation of sampling algorithms based on Wasserstein gradient flows (WGF) and forward-only diffusion processes (… | ↗Not briefed |
hn hn-ai | 1mo | The Daily Wire Is Quietly Shooting a Pro-ICE Movie. We Got the Script | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Winamp-style native audio player for macOS | Hi All - I've been missing Winamp for a long time and could not find a simple, fast audio player for macOS that I liked.<p>So, I… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LLM2Jev: LLMs Are Already Jev-Style Decision Models -- When and How to Fine-Tune Them | Jev-style decision models return categorical probability distributions over predefined options without generating free-form text, enab… | ↗Not briefed |
hn hn-ai | 1mo | macOS 27 is so buggy | — | ↗Not briefed |
rss techmeme | 1mo | Sony launches Quick Spectral Super Resolution, a "new performance tier of AI upscaling" for the standard PS5 developed in collaboration with AMD (Jay Peters/The Verge) | Jay Peters / The Verge: Sony launches Quick Spectral Super Resolution, a “new performance tier of AI upscaling” for the standard PS5 d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | what's the most pathetic thing you've ever asked an LLM? mine was "tips on how to get washing machine out from under the counter", despite… | what's the most pathetic thing you've ever asked an LLM? mine was "tips on how to get washing machine out from under the counter", de… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Homomorphic Advantage Operator: Stabilizing Reinforcement Learning Under Fully Homomorphic Encryption Constraints | Privacy-preserving machine learning presents significant deployment challenges on the cloud for intelligent systems with confidential… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly refreshing to see a recomp clearly stating this. Go check it out, tbh | Honestly refreshing to see a recomp clearly stating this. Go check it out, tbh | ↗Not briefed |