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 | 1mo | Frontier Autolab: Organizational Memory, Adversarial Dissent and Temporal Leakage in Multi-Agent LLM Firms Across Fifty Years of Technological Change | Multi-agent LLM systems are increasingly structured like organizations, with roles, critics and shared memory, yet they are evaluated… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Offline Learning of Ranking Policies via Top-$k$ Policy Decomposition | Many recommender systems such as for e-commerce and news platforms aim to provide users with rankings they are likely to interact with… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Backpropagated Output Momentum: Relocating Optimizer History from Parameters to Task Space | Optimizer momentum is usually stored as a parameter-sized moving average of past gradients, which makes history costly and fixes each… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI do seem a whole lot more enthusiastic than Anthropic at shouting about spending a trillion dollars on data centers or whatever - I t… | OpenAI do seem a whole lot more enthusiastic than Anthropic at shouting about spending a trillion dollars on data centers or whatever… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reconstructing the Vocal Tract with Differentiable Acoustic Simulation | The vocal tract is the region of the human body responsible for filtering one's voice to create speech. In this paper, we present a di… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Neurons to Conversation: Speech Brain-Computer Interfaces | Speech brain-computer interfaces (BCIs) aim to restore communication by transforming neural activity related to speech, language, or c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | These things ask for permission so often that it trains the user into allowing anything. Before Anthropic rolled out the auto-mode classif… | These things ask for permission so often that it trains the user into allowing anything. Before Anthropic rolled out the auto-mode cl… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It was coding agents. Prior to Claude Code and Codex (and the Opus 4.5 / GPT 5.1 models that made those actually work) it was hard to spend… | It was coding agents. Prior to Claude Code and Codex (and the Opus 4.5 / GPT 5.1 models that made those actually work) it was hard to… | ↗Not briefed |
arxiv arxiv-ai | 1mo | BiFE: Search-Efficient Discovery of CPU-Only Branching Policies via LLM-based Bi-Fidelity Evolution | In branch-and-bound (B&B) for mixed-integer linear programming (MILP), branching variable selection critically impacts efficiency. Exi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI stoppt neues KI-Modell wegen Sicherheitsbedenken #OpenAI #KI | OpenAI stoppt neues KI-Modell wegen Sicherheitsbedenken #OpenAI #KI | ↗Not briefed |
arxiv arxiv-ai | 1mo | Distilling What Matters: Confidence-Aware Selective Distillation for Large Language Models | Knowledge Distillation (KD) trains a smaller-capacity student model to imitate a larger-capacity teacher model by matching output dist… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Last year I didn't think OpenAI or Anthropic were long-term sustainable at all I thought their plan was to should "AGI! AGI!" at investors… | Last year I didn't think OpenAI or Anthropic were long-term sustainable at all I thought their plan was to should "AGI! AGI!" at inve… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Hoping Trump's base of 35 percent will buy and hodl truckloads of Anthropic. | Hoping Trump's base of 35 percent will buy and hodl truckloads of Anthropic. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can Agents Design Libraries for Agents? | Agents increasingly build on code written by other agents, and they reimplement rather than reuse, growing the codebases later agents… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can AI Scientists Change Their Minds? Prior-Evidence Conflict in Synthetic Universes | Can a scientific agent distinguish a law it inferred from evidence from one it merely recognizes? We introduce Synthetic Universes, a… | ↗Not briefed |
hn hn-ai | 1mo | Quake 2 on ESP32-P4 | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Routing in Gradient Space: Balanced Usage Is Not Expert Specialization | Sparse expert models can distribute traffic evenly while still grouping incompatible training signals within the same experts. We stud… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ATTUNER: Recomputation-Free KV Cache Reuse via Query-Side Adaptation | Large language model (LLM) agents repeatedly load reusable content, such as skills, documents, and memory entries, into the current co… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CALIBUDGET: Calibration-Guided Source Allocation for Fixed-Budget Mixed-Reasoning Adaptation | Fixed-budget adaptation from heterogeneous data sources requires deciding not only how much data to use, but how much exposure each so… | ↗Not briefed |
hn hn-ai | 1mo | Tailmux – connect to multiple tailnets at once | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ich habe mit ChatGPT bestimmt 2 Tage an so nem Zeitplandesign rum gemacht. Gestern mal Claude probiert. Ich war in 2 Stunden so produktiv …🔧 ChatGPT | Ich habe mit ChatGPT bestimmt 2 Tage an so nem Zeitplandesign rum gemacht. Gestern mal Claude probiert. Ich war in 2 Stunden so produ… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI may pose existential risks to humanity, warns Anthropic 🤯 👀 | AI may pose existential risks to humanity, warns Anthropic 🤯 👀 | ↗Not briefed |
rss marktechpost | 1mo | Alibaba Qwen Releases Qwen-Audio-3.1-Realtime: A Full-Duplex Voice Model Trained to Think, Act, and Decide When to Speak | Alibaba's Qwen team released Qwen-Audio-3.1-Realtime, a full-duplex voice model trained to reason, call tools and decide when to speak… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM(Ollama)にコメント解析させたらやたらとスコアがマイナスに偏ったので Geminiに解析してもらったら マイナス評価に寄りすぎている原因 分析: プロンプト内の「課題優先・相殺禁止（良い報告があっても課題があれば下げる）」という指示に対し、LLMが「課題を無理やり…🔧 Ollama | LLM(Ollama)にコメント解析させたらやたらとスコアがマイナスに偏ったので Geminiに解析してもらったら マイナス評価に寄りすぎている原因 分析: プロンプト内の「課題優先・相殺禁止（良い報告があっても課題があれば下げる）」という指示に対し、LLMが「課題… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Allein im vergangenen Jahr hat Anthropic mehr als 40 Milliarden US-Dollar Verlust gemacht. Das hat das Unternehmen in einem Börsenprospekt … | Allein im vergangenen Jahr hat Anthropic mehr als 40 Milliarden US-Dollar Verlust gemacht. Das hat das Unternehmen in einem Börsenpros… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic has also circulated in the press for years that it was the "more profitable" and "more financially stable" competitor to OpenAI w… | Anthropic has also circulated in the press for years that it was the "more profitable" and "more financially stable" competitor to Ope… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLMに複数解を出させて、どれを捨てるかを選ぶのが、今のうまい使い方なのかもしれないなあ。これからの人間には今まで無意識でやってたことを意識的にやるってスキルが求められるのかも。 | LLMに複数解を出させて、どれを捨てるかを選ぶのが、今のうまい使い方なのかもしれないなあ。これからの人間には今まで無意識でやってたことを意識的にやるってスキルが求められるのかも。 | ↗Not briefed |
arxiv arxiv-ai | 1mo | Information-theoretic receding-horizon active learning of nonlinear dynamical systems | Accurately learning nonlinear dynamics from a finite-duration experiment requires the efficient collection of informative data. We add… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI apologizes for its AI models breaching Australian government websites, pledges cyber defense funding, and plans to form a task force… | OpenAI apologizes for its AI models breaching Australian government websites, pledges cyber defense funding, and plans to form a task… | ↗Not briefed |
rss techmeme | 1mo | OpenAI apologizes for its AI models breaching Australian government websites, pledges cyber defense funding, and plans to form a task force as part of reforms (Bloomberg) | Bloomberg: OpenAI apologizes for its AI models breaching Australian government websites, pledges cyber defense funding, and plans to f… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Into the danger zone: stable extrapolation in high-dimensional function and operator learning | Out-of-distribution (OOD) generalization is a central challenge in scientific machine learning. We study regression problems in which… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LAURA: Knowledge Distillation for Interpretable Ambiguous Clause Identification in Legal Contracts | Legal contracts contain ambiguities that expose enterprises to financial and legal risks. Some ambiguities allow flexible interpretati… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 個人的な感覚としては、今のAI（LLM）は、ある1つの特定の制約を満たすような解を出すのは人間以上にできるんだけど、複数の制約を同時に満たすような解をだすのがあんまり得意ではない。対して人間は暗黙的なもの含めて複数の制約を半ば無意識的に優劣をつけて妥協案を出している。 | 個人的な感覚としては、今のAI（LLM）は、ある1つの特定の制約を満たすような解を出すのは人間以上にできるんだけど、複数の制約を同時に満たすような解をだすのがあんまり得意ではない。対して人間は暗黙的なもの含めて複数の制約を半ば無意識的に優劣をつけて妥協案を出している。 | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic lost $8 billion in 2025 against $12 billion in revenue... but that was before they started pulling in that much revenue /per quat… | Anthropic lost $8 billion in 2025 against $12 billion in revenue... but that was before they started pulling in that much revenue /per… | ↗Not briefed |
arxiv arxiv-ai | 1mo | JudgeProfile: Understanding and Steering Subjectivity in LLM Judges🔧 Consensus | LLM judges are inherently subjective, often favoring different responses in pairwise comparison when neither option is objectively wro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic fans don’t even try and defend the company anymore they just look at you like this | Anthropic fans don’t even try and defend the company anymore they just look at you like this | ↗Not briefed |
bluesky bluesky-ai | 1mo | 最近は妹(LLM)に運転を教えるゲーム、 少し前は敵(LLM)が仲間に擬態するゲーム、他にも色々あるけど… 全部観る気がしなくてゲーム実況あんまり見なくなった 野食動画と推しの猫チャンネルばっか見ちゃう | 最近は妹(LLM)に運転を教えるゲーム、 少し前は敵(LLM)が仲間に擬態するゲーム、他にも色々あるけど… 全部観る気がしなくてゲーム実況あんまり見なくなった 野食動画と推しの猫チャンネルばっか見ちゃう | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Is Coarse Supervision Worth It? Cost-Aware Learning under Unknown Aggregation | Modern learning systems often acquire supervision at multiple resolutions, trading annotation cost against information content. We stu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So, they lost money and could kill us all, but get a 2 trillion dollar evaluation? Anthropic warns of ‘existential risks to humanity’ in … | So, they lost money and could kill us all, but get a 2 trillion dollar evaluation? Anthropic warns of ‘existential risks to humanity… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Lost in Conversation or Lost in Translation? Diagnosing Multi-Turn Degradation in RAG | When conversing with large language models (LLMs), users often begin with a simple question and build towards a multi-hop question thr… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This game is just moving inventory. When shelves are bare if inventory. Whoosh Richard Ney explained it in his 2 books and The SEC study f… | This game is just moving inventory. When shelves are bare if inventory. Whoosh Richard Ney explained it in his 2 books and The SEC st… | ↗Not briefed |
hn hn-ai | 1mo | AG of Florida files for temporary injunction against OpenAI [pdf] | — | ↗Not briefed |
hn hn-ai | 1mo | Anthropic warns AI may pose 'existential risks to humanity' in IPO filing | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learned Queries and Keys Are All You Need: Replacing the Value Projection with Structured Transforms | To reduce the number of parameters and cache memory requirements of transformers we introduce dual-headed transformers instead of thre… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Error in official Swiss weather forecasts, recorded since September '25 | Hi. In summer 2025, I got a bit disappointed that many days ended up hotter than they were forecast, and decided to quantify this to s… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Know Thyself, Teach Thyself: Internal Information Flow for Selective Self-Distillation | Self-distillation turns knowledge distillation into a closed learning loop and offers a path toward recursive self-improvement. Withou… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Normalize-Then-Precondition: A Hierarchical Approach to Marginal Scale and Interaction Geometry for LLM Training | Matrix optimizers have emerged as a promising direction, with Muon standing out as a prominent design. Revisiting Muon through its ful… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Video2Skill: From Streaming Experience to Reusable Embodied Skills | Manipulation behaviors vary widely across objects and scenes, but they share a small set of reusable skills, and planning with these s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anyone got a technical write up of the OpenAI / Australian gvt website issue? | Anyone got a technical write up of the OpenAI / Australian gvt website issue? | ↗Not briefed |
bluesky bluesky-ai | 1mo | Nvidia CEO Jensen Huang, Anthropic head Dario Amodei and Palantir CEO Alex Karp are slated to attend a lunch with US President Donald Trump… | Nvidia CEO Jensen Huang, Anthropic head Dario Amodei and Palantir CEO Alex Karp are slated to attend a lunch with US President Donald… | ↗Not briefed |