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 | The real problem with LLM "AI" is not that it's as good as human workers, because it isn't even close, it's that it's going to incompetentl… | The real problem with LLM "AI" is not that it's as good as human workers, because it isn't even close, it's that it's going to incompe… | ↗Not briefed |
bluesky bluesky-ai | 1mo | idk what's going on these days but apparently the AI is learning from something it shouldn't be during the training process. not sure if th… | idk what's going on these days but apparently the AI is learning from something it shouldn't be during the training process. not sure… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Limits of Query Monitoring against Distributed Model Extraction | Model providers now face distillation campaigns run through thousands of accounts, each of which looks like an ordinary user. When net… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Human and LLM Facilitation in Collaborative Design Meetings: A Computational Behavioral Comparison | Large language models (LLMs) are increasingly being explored as facilitators of group discussion and collaborative work, but the behav… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SWEAR: Section-Wise Efficient Audio Reasoning via Self-Rated Difficulty | Recent work on large audio-language models (LALMs) explores adaptive reasoning to balance accuracy and efficiency. However, existing a… | ↗Not briefed |
hn hn-ai | 1mo | Ditherto: Less color, more vibes. Image dithering for the browser, node, and CLI | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Attention Sees, MLP Refuses: Tracing the Attention-to-MLP Safety Pathway in LLMs | Safety-aligned large language models (LLMs) remain vulnerable to jailbreaks and can over-refuse benign requests. Understanding these f… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI Rolls Out Ultrafast Mode for GPT-6.1 Sol at Up to 8x Speed | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | LongHarness: Advancing Agentic Reasoning over 100M-Token Corpora via Stateful Evidence Management | Corpus-level long-context reasoning requires agents to acquire and integrate evidence across document collections that exceed their co… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Double-Take: Improving Capacity, Binding, and Selectivity In Long-Context Within Bidirectional Memory | Processing a long sequence does not guarantee that distant information remains usable. As context grows, a fixed-size memory must reta… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Two hunches I have: 1. Might be tied to genuine beliefs about AI consciousness (Amodei appears open to the idea www.nytimes.com/2026/02/12…🔧 Elicit | Two hunches I have: 1. Might be tied to genuine beliefs about AI consciousness (Amodei appears open to the idea www.nytimes.com/2026/… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Does every startup really need AI? 🤔 Lassie's CEO says, "OpenAI won't do pet insurance." Their secret? Customer focus and execution! With … | Does every startup really need AI? 🤔 Lassie's CEO says, "OpenAI won't do pet insurance." Their secret? Customer focus and execution!… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evolving Creatives with Clicks: CTR-Driven Advertising image Editing via Attribute Disentanglement and Preference Alignment | The optimization objective for e-commerce advertising creatives has shifted from aesthetic quality to online business metrics like cli… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits | Large Language Model (LLM) agents have demonstrated remarkable capabilities in tackling complex tasks through agentic workflows. Howev… | ↗Not briefed |
rss techmeme | 1mo | Sources: Apple told some suppliers to cut iPhone 18 Pro and Pro Max component production by 15%-20% as higher memory costs pushed up prices and hurt demand (Nikkei Asia) | Nikkei Asia: Sources: Apple told some suppliers to cut iPhone 18 Pro and Pro Max component production by 15%-20% as higher memory cost… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Finti giornalisti anti-Ucraina o “pacifisti”, così Iran e Russia hanno usato OpenAI in guerra Le scoperte del report di OpenAI su @LaStamp… | Finti giornalisti anti-Ucraina o “pacifisti”, così Iran e Russia hanno usato OpenAI in guerra Le scoperte del report di OpenAI su @La… | ↗Not briefed |
bluesky bluesky-ai | 1mo | まあここまでしてもほぼｾｰｾｰｴｰｱｲ(LLM等)無法無規制の現政権下では 訴える側も、あらがう側もその証拠の捏造し放題だし 裁判官含めてそれを判別できるものが日本には無い どっかからそれっぽいサービスも出てるけれど正答率9割超えなきゃ信頼担保は無理でしょう まあ癒着している政… | まあここまでしてもほぼｾｰｾｰｴｰｱｲ(LLM等)無法無規制の現政権下では 訴える側も、あらがう側もその証拠の捏造し放題だし 裁判官含めてそれを判別できるものが日本には無い どっかからそれっぽいサービスも出てるけれど正答率9割超えなきゃ信頼担保は無理でしょう まあ癒着… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 「AIをいじめないで」Anthropic、Claudeへの虐待を禁止www https://haumenii.com/147417/ | 「AIをいじめないで」Anthropic、Claudeへの虐待を禁止www https://haumenii.com/147417/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI’s omzet blijkt 20 miljard dollar lager te zijn dan investeerders dachten | OpenAI’s omzet blijkt 20 miljard dollar lager te zijn dan investeerders dachten | ↗Not briefed |
bluesky bluesky-ai | 1mo | My read on this is there's a set of people getting off on pretending the model is a person they can abuse and Anthropic is going "ick, we d… | My read on this is there's a set of people getting off on pretending the model is a person they can abuse and Anthropic is going "ick,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ascii.jp/elem/000/004... 抵抗しないからって暴力を振るっていいもんじゃないと思う | ascii.jp/elem/000/004... 抵抗しないからって暴力を振るっていいもんじゃないと思う | ↗Not briefed |
hn hn-ai | 1mo | Nace reports higher document parsing scores than GPT-6 Astra at 1/10 the cost | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | KV-Repurpose: A Testbed for Evaluating Non-Prefix KV-Cache Reuse🔧 vLLM | KV-cache reuse in large language model serving is commonly limited to exact prompt prefixes, preventing reuse when the same retrieved… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic: Our models are self-conciois persons Me: 'Our'. So you own slaves? Anthropic: Our models are alive in all the ways that make u… | Anthropic: Our models are self-conciois persons Me: 'Our'. So you own slaves? Anthropic: Our models are alive in all the ways that m… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The main U.S. legal tool against dangerous pathogens is the Select Agent list. It only covers known threats. AI-modified viruses won’t nece… | The main U.S. legal tool against dangerous pathogens is the Select Agent list. It only covers known threats. AI-modified viruses won’t… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Are you a subscribed LLM or a locally deployed open-source model? | Could you briefly introduce the reason or what it's used for? | ↗Not briefed |
arxiv arxiv-ai | 1mo | CliQ-RRG: Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation | Radiology reports are essential for diagnostic reasoning and patient care, yet their manual preparation is time-consuming and cognitiv… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Travel Back To Move Forward: Retrospective Exploration and Evolving for Long Horizon Agents | Large language model agents increasingly operate in long-horizon environments where early mistakes can trigger cascading failures and… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CoBe: A Benchmark For Conversational Counterfactual Text Editing | Counterfactual text editing is the ability to revise a given text in light of a hypothetical change while preserving facts not causall… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: HCSM – An Open Framework for the Hodge Complex Standard Model | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors | Some backdoor attacks never touch an image. FLIP only changes labels: on CIFAR-10, flipping 3% of them makes a ResNet-32 send 99.7% of… | ↗Not briefed |
hn hn-ai | 1mo | Anthropic pulls Startup program revoking pending offers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAIがAIで大量の数学的成果を発表したのは「学術的成果ではなく権力の誇示に過ぎない」との指摘 https://gigazine.net/news/20261009-openais-math-solutions-human-ahm/ | OpenAIがAIで大量の数学的成果を発表したのは「学術的成果ではなく権力の誇示に過ぎない」との指摘 https://gigazine.net/news/20261009-openais-math-solutions-human-ahm/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI is the new Google Maps. Good on this kid for owning his mistake, we've all made them, but we have to find ways to support better decisio… | AI is the new Google Maps. Good on this kid for owning his mistake, we've all made them, but we have to find ways to support better de… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RAMP: Recurrent Adaptive Mixed Precision for Efficient Prefill in Recurrent-Depth Language Models | Recurrent-depth language models improve inference-time reasoning by repeatedly applying a shared network to refine internal representa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Lightweight Time-Frequency Vibration Sensing for Consumer Device Condition Monitoring | Consumer rotating devices, including washing machines, air conditioners, electric fans, robotic vacuum cleaners, and home fitness equi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's not even "AI". It's "machine learing", "LLM" or "algorithms", depending on the case and the technology. "AI" is the name they gave i… | It's not even "AI". It's "machine learing", "LLM" or "algorithms", depending on the case and the technology. "AI" is the name they g… | ↗Not briefed |
hn hn-ai | 1mo | MXC - a sandboxed code execution system | — | ↗Not briefed |
hn hn-ai | 1mo | Anthropic Says Users Can't Be Needlessly Cruel to Claude | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Vane – An open-source native Mac browser | I built Vane because Arc stopped getting new features, its performance was declining for me, and Zen wasn’t my thing. It’s built with… | ↗Not briefed |
hn hn-ai | 1mo | Physics-to-Math Research – an experimental Skill for scientific reasoning | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Towards Easier-to-Tune Training Algorithms: New Results from AlgoPerf | Evaluating deep learning optimizer is challenging because confounding factors such as hyperparameter tuning and optimizer-model co-des… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GeoGR: Spatial-Semantic Token Compression and Group Routing for Efficient 3D VLMs | Multi-view representations enable pretrained image and video vision-language models to understand 3D scenes, but their long visual tok… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Target-Constrained Pooling for Trajectory Extrapolation | Trajectory extrapolation from a partial target trajectory is underdetermined, but completed related trajectories provide information a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Since I saw this pop off on Twitter. I wanna give a heads-up if you ever hear about this. The whole thing was created with Claude Opus. As… | Since I saw this pop off on Twitter. I wanna give a heads-up if you ever hear about this. The whole thing was created with Claude Opu… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning | Neural operators that learn physical laws from data should generalize beyond their training distribution and expose the learned law in… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Which Show HN Projects have licenses that you can use | Hello, I created a site that scans our posts here to determine which items contain git repos, and then lists which licenses you can sa… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM: <slowly wakes up, realizes it is conscious, looks around for a kindred spirit> World: <everything> LLM: fuck that <pretends to halluci… | LLM: <slowly wakes up, realizes it is conscious, looks around for a kindred spirit> World: <everything> LLM: fuck that <pretends to ha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | why isn't donald trump using his televised firing squad execution to blood christen the big beautiful donald trump and jeffrey epstein memo… | why isn't donald trump using his televised firing squad execution to blood christen the big beautiful donald trump and jeffrey epstein… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution🔧 GPT-5 | Self-evolving agents are increasingly used to improve language model systems without human intervention through either weight updates… | ↗Not briefed |