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 | では、新たなキャラクターが登場です。 駅のホーム。 データに記録されていた被験体の一人——アキと名乗る人物が現れた。 「PROJECT LUXはまだ動いている」という警告。 しかし白夜は感じていた。 この接触自体が、組織の罠ではないか。アキの眼差しに隠された意図。 紗奈… | では、新たなキャラクターが登場です。 駅のホーム。 データに記録されていた被験体の一人——アキと名乗る人物が現れた。 「PROJECT LUXはまだ動いている」という警告。 しかし白夜は感じていた。 この接触自体が、組織の罠ではないか。アキの眼差しに隠された意図… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Does Muon Need Fine-Grained Spectral Shaping? | Muon combines current and past gradients into matrix momentum. For $M=UΣV^\top$, the idealized polar update $Q=UV^\top$ gives every si… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Even Musk’s chatbot was reacting to Kennedy with the AI equivalent of AYFKM." | "Even Musk’s chatbot was reacting to Kennedy with the AI equivalent of AYFKM." | ↗Not briefed |
bluesky bluesky-ai | 1mo | At day job today team talked about how out of date the LLM agent is that was built for us... Six months ago. Cannot even convey how fast … | At day job today team talked about how out of date the LLM agent is that was built for us... Six months ago. Cannot even convey how… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Behold. Maple Nekokami, v1 vtuber. This took several million tokens, harnesses, and several AI model providers (OpenAI and 11labs) | Behold. Maple Nekokami, v1 vtuber. This took several million tokens, harnesses, and several AI model providers (OpenAI and 11labs) | ↗Not briefed |
bluesky bluesky-ai | 1mo | I'm an economist who's spent much of the past decade conducting and reading research on this subject. Seriously, what the hell is this? | I'm an economist who's spent much of the past decade conducting and reading research on this subject. Seriously, what the hell is this? | ↗Not briefed |
bluesky bluesky-ai | 1mo | 名作絵本『がまくんとかえるくん』風にAIの暴走によるハッキングを風刺した『がまくんとかえるくんとますます進化する機械たち』 gigazine.net/news/2026100... | 名作絵本『がまくんとかえるくん』風にAIの暴走によるハッキングを風刺した『がまくんとかえるくんとますます進化する機械たち』 gigazine.net/news/2026100... | ↗Not briefed |
arxiv arxiv-ai | 1mo | Who Bears the Burden? Learning Responsibility for Shared Constraints in Multi-Agent Reinforcement Learning | When multiple agents share a cost budget, a common Lagrange multiplier can enforce the aggregate constraint but does not determine how… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 今朝配信の結城メルマガ Vol.758 を読みました。 いちばん心に残ったのは、数学は「理解するまではさっぱりわからない。理解するとあたりまえに思える」という話です。わからないうちは書けないし、わかってしまうとあたりまえすぎて書けない。自転車の乗り方を文章で説明する難しさにた… | 今朝配信の結城メルマガ Vol.758 を読みました。 いちばん心に残ったのは、数学は「理解するまではさっぱりわからない。理解するとあたりまえに思える」という話です。わからないうちは書けないし、わかってしまうとあたりまえすぎて書けない。自転車の乗り方を文章で説明する… | ↗Not briefed |
bluesky bluesky-ai | 1mo | How should I respond when a psychopath tells me I should expect bad things to happen? www.the-independent.com/tech/ai-sam-... | How should I respond when a psychopath tells me I should expect bad things to happen? www.the-independent.com/tech/ai-sam-... | ↗Not briefed |
arxiv arxiv-ai | 1mo | Deep Defence on Wheels: A Dual Intrusion Detection System Architecture for Comprehensive In-Vehicle Network Security | Increasing connectivity to the outside world and the lack of inbuilt security mechanisms have made legacy intra-vehicular networks vul… | ↗Not briefed |
rss marktechpost | 1mo | Meet Together Link: A Free CLI That Runs Open Models Like Kimi K3 and GLM 5.3 Inside Claude Code, Codex, and OpenCode🔧 Together AI🔧 ChatGPT | Together AI has released Together Link, a free, MIT-licensed CLI that connects Claude Code, Codex, OpenCode, Pi, and the Claude and Ch… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Robust Importance Sampling for Rare Events via Constrained Gaussian Mixtures | We study estimating rare-event probabilities $I = \mathbb{P}(g(\mathbf{X}) > γ)$ with $\mathbf{X} \sim \mathcal{N}(\boldsymbolμ, \bold… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SpecBraM: What Should an EEG Foundation Model Predict? Masked Band-Power Prediction versus Waveform Reconstruction | Self-supervised EEG models often reconstruct masked waveforms or predict discrete codes. We study a task-aligned alternative: masked b… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A lightweight mapping editor for improving MTB trails on OpenStreetMap | I wanted to share a lightweight mapping editor I’ve built to make it easier to improve MTB trail data on OpenStreetMap. In general MTB… | ↗Not briefed |
arxiv arxiv-ai | 1mo | In With the Old: Enhancing 'Classical' Document Automation with Generative AI | Software-based legal assistance systems have leveraged many different forms of knowledge representation and reasoning. This article ex… | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.theguardian.com/technology/2... Shockingly immoral tech arseholes | www.theguardian.com/technology/2... Shockingly immoral tech arseholes | ↗Not briefed |
bluesky bluesky-ai | 1mo | If paying for their training data via a scheme such as this means they can't afford as many GPUs and data centers, you know, I think that t… | If paying for their training data via a scheme such as this means they can't afford as many GPUs and data centers, you know, I think t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can Power Draw Constrain Covert Compute? Limits of Analogue Verification for AI Governance | Frontier AI treaties or agreements on limiting computation require external verification; an external auditor must be able to confirm… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Adapting to Changes in Agent Behavior via Finite-Depth Policy Sensitivity | Adapting a reinforcement learning policy to changes in another agent's behavior typically requires a large amount of new interaction d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PsyCIDRA: A Dual-Agent Framework for Psychiatric Interviewing and Diagnostic Reasoning | Large language models show promise in clinical reasoning, but psychiatric interviewing requires guiding an evolving conversation. Thei… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 'Freaking Insane': Sam Altman Sparks Horror by Saying We Should 'Accept Some Bad Things Happening' for Sake of AI I remember when we used… | 'Freaking Insane': Sam Altman Sparks Horror by Saying We Should 'Accept Some Bad Things Happening' for Sake of AI I remember when we… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I really, really, really should write up the "mechanical license for LLM training" decision brief that I've been talking about for ages | I really, really, really should write up the "mechanical license for LLM training" decision brief that I've been talking about for ages | ↗Not briefed |
bluesky bluesky-ai | 1mo | one of the implicit premises in a lot of pro-LLM consciousness arguments, which i hold is actually false, is the assumption that there is s… | one of the implicit premises in a lot of pro-LLM consciousness arguments, which i hold is actually false, is the assumption that there… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Structure, Not Belief: Correlated Thompson Sampling from LLM-Derived Covariance in Combinatorial Semi-Bandits | Combinatorial Thompson sampling (CTS) draws independent posterior samples for every arm, so its exploration dynamics ignore any relati… | ↗Not briefed |
arxiv arxiv-ai | 1mo | COMPASS: Finding Where Reasoning Lives in Language Models | Explicitly eliciting reasoning substantially improves LLM performance. Existing approaches require a predefined characterization of re… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Sam Altman of OpenAI said, very heartfelt-tedly, 'if we don't get to use people's content for free, we cannot afford to make this world-ch… | "Sam Altman of OpenAI said, very heartfelt-tedly, 'if we don't get to use people's content for free, we cannot afford to make this wor… | ↗Not briefed |
hn hn-ai | 1mo | Nuk4sd Sandbox Vault System | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think there's a pretty good argument that training an LLM is transformative and should be allowed under fair use (although inference uses… | I think there's a pretty good argument that training an LLM is transformative and should be allowed under fair use (although inference… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The federal AI hearing getting underway now. ABC up first. OpenAI/Anthropic et al are later today. Streaming here. www.youtube.com/watch?v=… | The federal AI hearing getting underway now. ABC up first. OpenAI/Anthropic et al are later today. Streaming here. www.youtube.com/wat… | ↗Not briefed |
hn hn-ai | 1mo | Learning Jazz Pianist Style with Cross-Attention Conditioning | — | ↗Not briefed |
rss arstechnica-tech | 1mo | MCP for agent-to-agent comms may be the riskiest protocol you've never heard of | Trust gaps in the new protocol spread malicious prompts from one agent to another. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Multimodal Inference through Adaptive Acquisition and Sequential Fusion | Multimodal systems often encode every available input, even when a subset suffices for prediction. Adaptive acquisition can reduce thi… | ↗Not briefed |
hn hn-ai | 1mo | Better Space-Time Trade-Offs for LSM-Tree Based Key-Value Stores [pdf] | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | What? This is publicly stated info with news articles and statements from the Taylor's themselves. How's is this contentious? Read your lin… | What? This is publicly stated info with news articles and statements from the Taylor's themselves. How's is this contentious? Read you… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "In an interview on Monday, OpenAI CEO Sam Altman told Politico that the world 'should accept some bad things happening for the benefits of… | "In an interview on Monday, OpenAI CEO Sam Altman told Politico that the world 'should accept some bad things happening for the benefi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | seems likely that it’s no mistake that this openai cost & stability sprint comes right after announcement of Anthropic’s IPO | seems likely that it’s no mistake that this openai cost & stability sprint comes right after announcement of Anthropic’s IPO | ↗Not briefed |
bluesky bluesky-ai | 1mo | Even if they magically had every scrap of writing and a transcript of every sentence ever uttered by anything that has ever spoken in this … | Even if they magically had every scrap of writing and a transcript of every sentence ever uttered by anything that has ever spoken in… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Shut up and sit down, Sammy, you creepy little man www.politico.com/news/2026/10... | Shut up and sit down, Sammy, you creepy little man www.politico.com/news/2026/10... | ↗Not briefed |
rss techmeme | 1mo | Analysis: Anthropic's subscriptions offer ~5x more API-equivalent value per month than OpenAI's for agentic workloads with Claude Opus 5.5 vs. GPT-6.1 Sol (SemiAnalysis) | SemiAnalysis: Analysis: Anthropic's subscriptions offer ~5x more API-equivalent value per month than OpenAI's for agentic workloads wi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Tagging in the folks who are the most closely watching the anthropic stuff to this sideshow... Yet another thing authors have to deal with … | Tagging in the folks who are the most closely watching the anthropic stuff to this sideshow... Yet another thing authors have to deal… | ↗Not briefed |
bluesky bluesky-ai | 1mo | There is not enough publicly available copyright-free material to make an LLM, particularly not one that will reflect current modern unders… | There is not enough publicly available copyright-free material to make an LLM, particularly not one that will reflect current modern u… | ↗Not briefed |
hn hn-ai | 1mo | Terminal Steps: A Claude mod for a daily step goal, synced from Apple Health | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | An addendum: when people argue about whether the field stuck to its 2022/2023 guns by positioning an LLM in a more complex system to struct… | An addendum: when people argue about whether the field stuck to its 2022/2023 guns by positioning an LLM in a more complex system to s… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ElasticFit: Fit-Aware 3D Object Insertion via VLM Reasoning and Generative Adaptation | Inserting objects into existing 3D scenes requires more than selecting a plausible location: the inserted object must also fit local g… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is absolutely horrible, and clearly they used some LLM program to look for LLM writing. I'm so sorry. | This is absolutely horrible, and clearly they used some LLM program to look for LLM writing. I'm so sorry. | ↗Not briefed |
bluesky bluesky-ai | 1mo | I mean, look! Good things soon! Really! Pinky swear! You need to accept the harms now because good things soon! www.politico.com/news/2026… | I mean, look! Good things soon! Really! Pinky swear! You need to accept the harms now because good things soon! www.politico.com/news… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Auditable Claims about AI Agents | Organizations make claims about their AI agents: a person approves every external email, every action is logged, an evaluation shows t… | ↗Not briefed |
hn hn-ai | 1mo | One year later: the power of 1.1.1.1 interns | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Interpretable Hypergraph Learning via Neural Additive Models | Hypergraphs offer a natural framework for modeling networked data, where dependencies among entities are governed by higher-order inte… | ↗Not briefed |