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 | you dont actually need to train on millions of movies, videos, and novels to build laundry bots (or to produce an entirely capable Claude C… | you dont actually need to train on millions of movies, videos, and novels to build laundry bots (or to produce an entirely capable Cla… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Rashomon – An independent execution record for AI coding agents | I have been working on Rashomon, an open-source execution recorder for AI coding agents.<p>The basic idea is that the agent transcript… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This newsletter has become increasingly common in recent years, promoted by venture investors and savvy transactional attorneys. www.bloomb… | This newsletter has become increasingly common in recent years, promoted by venture investors and savvy transactional attorneys. www.b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Clammy Sam" Altman out here doing his best Lord Farquaad impersonation. "Some of you may have your life's work stolen or develop 'AI' psyc… | "Clammy Sam" Altman out here doing his best Lord Farquaad impersonation. "Some of you may have your life's work stolen or develop 'AI'… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Altman names Macron as the world leader with the most impressive approach to artificial intelligence, praising his safety focus. The OpenAI… | Altman names Macron as the world leader with the most impressive approach to artificial intelligence, praising his safety focus. The O… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If only AI researcher spin could be harnessed and made into clean power. Perhaps we could build a large number of human-sized hamster whe… | If only AI researcher spin could be harnessed and made into clean power. Perhaps we could build a large number of human-sized hamste… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sam Altman says Elon Musk is a 'bully' who only respects people who punch him back https://bit.ly/3VSuxp3 | Sam Altman says Elon Musk is a 'bully' who only respects people who punch him back https://bit.ly/3VSuxp3 | ↗Not briefed |
hn hn-ai | 1mo | Sdrtop – Terminal monitor for SDR hardware | — | ↗Not briefed |
rss theverge-ai | 1mo | OpenAI is adding text watermarking in ChatGPT and Codex🔧 ChatGPT | An invisible, machine-readable watermark in text output is rolling out to ChatGPT and Codex, but only for users in the European Union… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Dinowitz asks a good ? about "A.i." directed at kids, mentions cognitive decline, cheating. OpenAI lady keeps saying "OpenAI takes [X] ver… | Dinowitz asks a good ? about "A.i." directed at kids, mentions cognitive decline, cheating. OpenAI lady keeps saying "OpenAI takes [X… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 👀 per CNBC: NYC AI hearing: *Anthropic whistleblower Jacob Coxon, who raised early concerns called out the industry’s “extremely reckle… | 👀 per CNBC: NYC AI hearing: *Anthropic whistleblower Jacob Coxon, who raised early concerns called out the industry’s “extremely r… | ↗Not briefed |
hn hn-ai | 1mo | The-engineering-behind-LLM-inference | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is EXACTLY what large language model (LLM) AIs are doing—and they consume huge amounts of power and resources to do so. When I was a … | This is EXACTLY what large language model (LLM) AIs are doing—and they consume huge amounts of power and resources to do so. When I w… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I was thinking about the Anthropic, SpaceX, and OpenAI IPOs. What this might be a harbinger of is the end of public markets as a mechanism … | I was thinking about the Anthropic, SpaceX, and OpenAI IPOs. What this might be a harbinger of is the end of public markets as a mecha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Not sure what to tell you, but LLM are literally “using ML in linguistics”! They evolved from earlier approaches in ML for natural language… | Not sure what to tell you, but LLM are literally “using ML in linguistics”! They evolved from earlier approaches in ML for natural lan… | ↗Not briefed |
arxiv arxiv-ai | 1mo | One Figure, Every Canvas: Editable Flowchart Relayout via Agentic Pipeline | Pipeline figures in ML papers must be repurposed across many canvases, including paper columns, 16:9 slides, portrait posters, 1:1 soc… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Base Models Can Reason By Taking a Cue From Training Data🔧 Elicit | In this paper, we study how training data creates associations between the tokens at the start of a base model's response and the reas… | ↗Not briefed |
arxiv arxiv-ai | 1mo | BiasFlow: Geometric Monitoring and Backbone Regularization for Spurious Feature Reliance | Worst-group accuracy (WGA) evaluates a trained predictor but does not characterize how its frozen backbone behaves when a new head is… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning to Read the Contextual Tokens in Diffusion Transformers | Multimodal Diffusion Transformers (MM-DiTs) jointly process visual and textual representations throughout generation. These models rep… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Recursive Video In-Context Learning for Agentic Robot | LLM agents that orchestrate frozen vision-language-action (VLA) policies improve across episodes through text memory, which records wh… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Direct Intermediate Initialization for Tilted Diffusion Samplers | Some diffusion posterior samplers construct Gaussian-tilted intermediate distributions along the reverse process. We observe that thes… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Towards Looped Models Done Right, Part II: Rethinking at Fixed Points | Every recurrence of a looped language model adds cost in training, decoding, prefill, and reinforcement learning (RL). The closer recu… | ↗Not briefed |
arxiv arxiv-ai | 1mo | UniSlider: Perceptually Uniform Sliders for Continuous Image Editing | Sliders provide an intuitive interface for continuous image editing. In current generative approaches, however, the slider is simply a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents | Memory has become integral to the LLM agent ecosystem, supporting information retention and reuse across interactions. However, most e… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling🔧 GPT-5 | Open-source web agents are now strong enough to execute realistic browser tasks, but training them with reinforcement learning still d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Seems like this scale would treat procedurally generated content (e.g. levels in roguelikes, world maps in Civilization) as the same as e.g… | Seems like this scale would treat procedurally generated content (e.g. levels in roguelikes, world maps in Civilization) as the same a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data | Scientific figures often encode quantitative results that are not readily available in machine-readable form, making accurate plot dig… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TasteVal: Measuring the Experimental Research Taste of AI Systems Against Human Experts | We introduce TasteVal, a benchmark to evaluate the experimental research taste of frontier models. We define research taste as the abi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Deep Learning for Sleep Heart Rate Estimation from Accelerometers: Toward Population-Scale Cardiac Insight Without Optical Sensors | Large longitudinal cohorts often contain wrist accelerometry without optical heart-rate sensing, motivating recovery of cardiac inform… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Private online learning and prediction for Littlestone classes | We study mistake bounds for differentially private online learning and online prediction under oblivious realisable adversaries. Onlin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | La polémique quand on va se rendre compte que c'est écrit par un LLM va être drôle | La polémique quand on va se rendre compte que c'est écrit par un LLM va être drôle | ↗Not briefed |
arxiv arxiv-ai | 1mo | Paradee: Distilling Kokoro-82M into an 8M-Parameter Single-Voice Text-to-Speech Model | We distill Kokoro-82M, a widely used open text-to-speech model with 54 voices, into Paradee, an 8.07M-parameter model that speaks one… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In case you needed a reminder that Claude isn’t a place to share deep dark secrets, here you go www.techspot.com/news/114091-... | In case you needed a reminder that Claude isn’t a place to share deep dark secrets, here you go www.techspot.com/news/114091-... | ↗Not briefed |
arxiv arxiv-ai | 1mo | TAPDreamer: Transferable Adversarial Patches for World Action Models | World models learn to predict how their environment will evolve, making them an important foundation for general-purpose robotic contr… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Postquel, a Free Mac Postgres Client with Claude Code and Codex | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Finding Gaussian Structure in Bosonic States | We study agnostic tomography of pure bosonic Gaussian states: given copies of an arbitrary $n$-mode bosonic state $ρ$, the goal is to… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Block Disentanglement in CRL: Bridging Identifiability and Visual State Estimation | Causal representation learning (CRL) is the process of recovering causally-related latent variables from high-dimensional observations… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI "rogue" agent activities found on Wikimedia projects | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | H-JEPA: End-to-End Learning of Hierarchical World Models for Visual Planning | Long-horizon planning with latent world models requires reasoning across timescales and levels of abstraction. Existing task-agnostic… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sharpen Without Search: On-Policy Distillation of Sequence-Level Power Distribution | A language model can give a correct answer more probability than any single incorrect answer and still usually sample an incorrect one… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI CEO Sam Altman is wrong — the world shouldn't 'accept some bad things' from AI: Zscaler CEO | OpenAI CEO Sam Altman is wrong — the world shouldn't 'accept some bad things' from AI: Zscaler CEO | ↗Not briefed |
bluesky bluesky-ai | 1mo | Secular game devs maybe, but religious game devs will use an Unreal 5.8 template with an LLM and saturate the marketplace with demoware lik… | Secular game devs maybe, but religious game devs will use an Unreal 5.8 template with an LLM and saturate the marketplace with demowar… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MC-Sparse: Deconstructing and Closing the Dense-Sparse Attention Gap in Diffusion Transformers | Sparse attention is a primary approach to reducing the latency of diffusion transformers in long-sequence generation tasks, such as vi… | ↗Not briefed |
hn hn-ai | 1mo | Cosmopolitan Libc Project Abandoned? | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Response Theory Probe for Learned Stochastic AI Simulators, Tested on Lorenz-63 | Machine-learning emulators of chaotic and stochastic systems are usually validated on forecast skill and long-run statistics. Neither… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wäre das Wasserzeichen robuster und schwerer zu entfernen, könnten Nutzer, die keine Erkennung wollen, zu Open-Weight-Modellen abwandern. S… | Wäre das Wasserzeichen robuster und schwerer zu entfernen, könnten Nutzer, die keine Erkennung wollen, zu Open-Weight-Modellen abwande… | ↗Not briefed |
rss techmeme | 1mo | Former Anthropic researcher Jacob Coxon and representatives from Anthropic, Google, OpenAI, and Meta testified at a New York City Council hearing on AI safety (Bloomberg) | Bloomberg: Former Anthropic researcher Jacob Coxon and representatives from Anthropic, Google, OpenAI, and Meta testified at a New Yor… | ↗Not briefed |
hn hn-ai | 1mo | Spec for Agent Memory Repo | — | ↗Not briefed |
hn hn-ai | 1mo | Sashiko: Agentic review of Linux kernel code | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | People are upset about llm/agent use for recomp projects, so I don’t think there much rationality behind it. | People are upset about llm/agent use for recomp projects, so I don’t think there much rationality behind it. | ↗Not briefed |