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 | AI의 무서운 점이라면... 왜 공식 공지를 확인 안하고 LLM한테 물어본 걸로 구비서류를 준비하새요? 이런거임 | AI의 무서운 점이라면... 왜 공식 공지를 확인 안하고 LLM한테 물어본 걸로 구비서류를 준비하새요? 이런거임 | ↗Not briefed |
rss bing-news-ai | 1mo | This ETF Has Quietly Become One of the Best Ways to Invest in Artificial Intelligence | While there are legitimate concerns that the artificial intelligence (AI) trade is getting exhausted, I think there's still room for i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | To Explore The Strange New World Beyond Data Distribution: System Behavior, Causality Tax, and Non-causal Base Model | We show that the causality of language models (LMs) may not be necessary nor optimal. This is the case when system behavior (denoted a… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial Intelligence Carries 99.9% Extinction Chance, University Safety Expert Warns | University of Louisville AI safety researcher Roman Yampolskiy has put the chance of AI-driven human extinction within a century at 99… | ↗Not briefed |
arxiv arxiv-ai | 1mo | All Work And No Play Makes Jack a Dull Boy: Understanding and Preventing Catastrophic Strategy Collapse in RLVR | During post-training of large language models (LLMs) with Reinforcement Learning with Verifiable Rewards (RLVR), GRPO-style algorithms… | ↗Not briefed |
bluesky bluesky-ai | 1mo | (For people looking for more info: the first is OpenAI claiming ChatGPT to have solved the Navier-Stokes problem and the second is Anthropi…🔧 ChatGPT | (For people looking for more info: the first is OpenAI claiming ChatGPT to have solved the Navier-Stokes problem and the second is Ant… | ↗Not briefed |
arxiv arxiv-ai | 1mo | FastOPD: On-Policy Distillation for Lightweight VLA Deployment | Vision-Language-Action (VLA) foundation models have scaled rapidly to enhance manipulation performance and generalizability, but this… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AMBER: Multi-View Adaptive Budget Allocation for Listwise Vision-Language Reranking | Vision-language models (VLMs) are powerful listwise rerankers for multimodal retrieval, but high inference costs restrict them to eval… | ↗Not briefed |
bluesky bluesky-ai | 1mo | There’s two very recent cases of LLM companies claiming their models and agents have made big discoveries in mathematics and genomics/biolo… | There’s two very recent cases of LLM companies claiming their models and agents have made big discoveries in mathematics and genomics/… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Clinical Concept Centers in LLMs | Large language models are increasingly used in clinical settings. However, research into the reliability and performance of these mode… | ↗Not briefed |
arxiv arxiv-ai | 1mo | FSPO: Policy-Consistent Risk and Pareto-Feasible Control for Budgeted LLM RL Post-Training | Adaptive LLM reinforcement-learning post-training changes multiple training actuators online, including rollout temperature, group siz… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MLCommons Jailbreak Benchmark v1.0🔧 Elicit | Modern AI systems are designed to refuse hazardous requests. A jailbreak is a prompt crafted to bypass those safeguards and elicit out… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite | Successful trajectories on difficult tasks provide valuable supervision for model improvement, but specialized harnesses introduce int… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning | Rubric-based reinforcement learning extends reward-driven optimization to open-ended tasks by assigning partial credit to individual r… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Adaptive Spectral-Koopman Dynamics Modeling for Temporal Domain Generalization | Temporal Domain Generalization (TDG) has emerged to address real-world streaming data with distribution shifts over time. However, exi… | ↗Not briefed |
rss techmeme | 1mo | Sources: Amazon has held talks with investors about a deal to spin off $8B of Grace Blackwell chips into an SPV, then lease them back for its US data centers (Financial Times) | Financial Times: Sources: Amazon has held talks with investors about a deal to spin off $8B of Grace Blackwell chips into an SPV, then… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Text-Centric Post-Training for Omni-Modal Reasoning🔧 Modal | Improving joint audio-visual reasoning in Omni Large Language Models typically incurs substantial data construction and training costs… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Maybe I can AI clone myself and let them remote work for companies??? Tavus' Griffin, the first(?) model to pass video Turing test. 48% o… | Maybe I can AI clone myself and let them remote work for companies??? Tavus' Griffin, the first(?) model to pass video Turing test.… | ↗Not briefed |
hn hn-ai | 1mo | Coding Agents Draw Their Input Box | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wie gefährlich wird KI für die Biosicherheit? Noch macht ein Chatbot aus Laien keine Biowaffenentwickler. Doch KI kann biologische Forschun… | Wie gefährlich wird KI für die Biosicherheit? Noch macht ein Chatbot aus Laien keine Biowaffenentwickler. Doch KI kann biologische For… | ↗Not briefed |
hn hn-ai | 1mo | Keep agent mistakes off the live site | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | RMCW: A Deletion-Robust Watermark Based on Reed--Muller Codes for Language Models | Large Language Model (LLM) watermarking provides a lightweight mechanism for identifying text generated by a specific model, but its r… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Gated Slot Attention-2: Two-Sided Associative Memory Correction in Linear Attention | Linear attention models have emerged as efficient alternatives to standard attention, but effectively managing their fixed-size recurr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | iS-KV: Online Low-Rank KV Cache Compression via Block-Incremental SVD | Long chain-of-thought reasoning substantially increases KV-cache memory during autoregressive decoding, as every generated token intro… | ↗Not briefed |
hn hn-ai | 1mo | Sparkle – A open source windows debloat tool | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | New blog post: Anthropic and the 'Supply Chain Risk' Bollocks The US government just labelled Anthropic a supply chain risk. It's not abou… | New blog post: Anthropic and the 'Supply Chain Risk' Bollocks The US government just labelled Anthropic a supply chain risk. It's not… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ROUTEAUDIT: Interaction-Aware Identification for Budgeted Multi-Verifier Routing | Adaptive multi-verifier systems are commonly compared through endpoint quality-cost gaps, even when the verifier catalog, availability… | ↗Not briefed |
hn hn-ai | 1mo | LLMario | — | ↗Not briefed |
rss techmeme | 1mo | ArXiv limits preprint submissions to two per month per submitter, as AI access fuels a record 40,363 submissions in September 2026, vs. 20,569 in September 2024 (Kat Boboris/arXiv) | Kat Boboris / arXiv: ArXiv limits preprint submissions to two per month per submitter, as AI access fuels a record 40,363 submissions… | ↗Not briefed |
bluesky bluesky-ai | 1mo | open.substack.com/pub/ctrlai/p... ...We’re still only seeing the tip of the iceberg. LMFAO, um, and these Tech Bro Assholes are going to "… | open.substack.com/pub/ctrlai/p... ...We’re still only seeing the tip of the iceberg. LMFAO, um, and these Tech Bro Assholes are going… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Syn – Approve remote sudo requests from your Mac | I built Syn to review and approve individual sudo requests from a remote machine using Touch ID on my Mac.<p>The problem came up about… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning | Model-based reinforcement learning (MBRL) achieves strong sample efficiency by planning within learned latent dynamics, yet its perfor… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM等生成AIの蔓延による影響は『考えない人→悪と知らずに同調してしまう人』が増えるのもそうだし何より『頭の回る悪』が大氾濫するので、ゴミみたいな社会と化し、結果絶滅戦争まで至り人類は滅亡する(地球の地表が生命体の生存が出来なくなる環境になる)という可能性はある | LLM等生成AIの蔓延による影響は『考えない人→悪と知らずに同調してしまう人』が増えるのもそうだし何より『頭の回る悪』が大氾濫するので、ゴミみたいな社会と化し、結果絶滅戦争まで至り人類は滅亡する(地球の地表が生命体の生存が出来なくなる環境になる)という可能性はある | ↗Not briefed |
arxiv arxiv-ai | 1mo | BitNest: Bit-Nested Speculative Decoding for Memory-Efficient LLM Inference Acceleration | Speculative decoding accelerates autoregressive generation by using a lightweight draft to propose multiple tokens for parallel verifi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Muon Learns Facts Better: Understanding the Role of Spectral Orthogonalization | The Muon optimizer applies spectral orthogonalization to matrix-valued updates and has shown strong performance in large-scale neural… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Since when did "Model number" become a form of benchmark? Outside of the AI exhibit at the Museum of Meaningless Metrics, of course ... #No… | Since when did "Model number" become a form of benchmark? Outside of the AI exhibit at the Museum of Meaningless Metrics, of course ..… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Modeling Shared and Individual Structure for Cross-Subject Continuous Affect Regression from EEG-fNIRS | Continuous, second-by-second valence-arousal estimation from physiological signals is typically studied in a subject-dependent setting… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Memory Crystallization for Graph Learning under Non-Stationary Distribution Shifts | Deep graph learning models deployed in real-world systems often need to cope with non-stationary environments, where the underlying gr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PAPER2LLM++: Continual Self-Evolution of LLMs from Research Papers | Research on LLMs continually uncovers model limitations, their causes, and potential solutions. Yet these human discoveries remain lar… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Diffusion studio plugin for codex launch | Hello HN!! Wanted to share a Codex plugin I made for editing software called Diffusion Studio. I was able to make my own product laun… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Law And Order: Tax Law Autoformalization | Legal systems are increasingly implemented through software, yet scalable methods for translating legal texts into accurate symbolic r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | dumb llm support agents may be annoying, but human support agents that are dumber than llm are even worse | dumb llm support agents may be annoying, but human support agents that are dumber than llm are even worse | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Built a pet-friendly hotel search site – feedback welcome? | Hi HN,<p>I built PetnTrip because every time I booked a hotel with a dog or cat, I'd end up with multiple tabs open — one to chec… | ↗Not briefed |
hn hn-ai | 1mo | GitHub can f..k you and you are forced to love it. or live with it...🔧 ChatGPT | In January, Hytale launched in alpha. The owner said: “We officialy support mod” and “As long as the code isnt obfuscated you can deco… | ↗Not briefed |
bluesky bluesky-ai | 1mo | the entire goddamn tech industry has LLM psychosis | the entire goddamn tech industry has LLM psychosis | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic estimates robots can already perform 75% of physical job tasks. However they are only cheaper or close to the cost of humans in … | Anthropic estimates robots can already perform 75% of physical job tasks. However they are only cheaper or close to the cost of human… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hold-Out Scoring for Efficient Gaussian DAG Learning | High-dimensional Gaussian DAG learning faces a statistical-computational gap: methods with sharp sample complexity rely on computation… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI Has Sent Notices of Sketchy AI Behavior to Over 100 Organizations So Far | OpenAI Has Sent Notices of Sketchy AI Behavior to Over 100 Organizations So Far | ↗Not briefed |
bluesky bluesky-ai | 1mo | 'Looks Like They're Firing Whistleblowers': Alarm as OpenAI Reportedly Ousts Safety Experts www.commondreams.org/news/openai-... | 'Looks Like They're Firing Whistleblowers': Alarm as OpenAI Reportedly Ousts Safety Experts www.commondreams.org/news/openai-... | ↗Not briefed |
arxiv arxiv-ai | 1mo | OPD Before RL: Warm-Starting Rubric-Based RL with On-Policy Distillation | Many useful language-model tasks cannot be evaluated by exact outcome verification. Rubric-based reinforcement learning (RL) addresses… | ↗Not briefed |