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 |
|---|---|---|---|---|
hn hn-ai | 1mo | ZQK – An open-core substrate for autonomous agent swarms in Go | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Large Stepsizes Federated Learning on Logistic Regression with Linearly Separable Data: The Case of Heterogeneous Devices | This paper revisits the distributed learning problem for training a multinomial logistic regression model with the Federated Averaging… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CoHyFuse: Condition-wise Hypergraph Fusion with Global Connectome in Task-fMRI | Task-fMRI connectomes reveal state-dependent neural reconfigurations, yet conventional methods marginalize these signals by aggregatin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MiniCorp: The Last Mile of the AI Agent Firm | The last mile toward enterprise AGI is a company that runs itself. Training and adapting such agents require longitudinal enterprise d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | David Robinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specific ru… | David Robinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specif… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It didn't work. | It didn't work. | ↗Not briefed |
arxiv arxiv-ai | 1mo | On Hyperparameter Tuning on the Test Set | "Don't tune hyperparameters on the test set" is often stated in machine learning textbooks. Violating it is considered a cardinal sin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Collaborative Personalized Preference Alignment for LLMs under Data Deficiency | Real-world users often exhibit highly heterogeneous preferences over multiple objectives for LLM responses. A lightweight aligner can… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Fitting Vision Adapters at Frontier Scales🔧 Kimi K2 | Training a small projector between a frozen vision encoder and language model is an established approach to multimodal learning. As th… | ↗Not briefed |
bluesky bluesky-ai | 1mo | sure, of course. true ai may be a possibility! but LLM procedural generation is not it and fundamentally can never be it. | sure, of course. true ai may be a possibility! but LLM procedural generation is not it and fundamentally can never be it. | ↗Not briefed |
arxiv arxiv-ai | 1mo | TasteRoute: Personalized Routing for Video Generation🔧 Consensus | Rapid progress in video generation has led to a plethora of models that differ substantially in capability and generation cost. This r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Πρέπει να επινοήσουμε μια νέα λέξη για τους κομπογιαννίτες της ΤΝ. Φτάσαμε στο σημείο να τους διορθώνει το Παπαδαριό. www.lifo.gr/now/world… | Πρέπει να επινοήσουμε μια νέα λέξη για τους κομπογιαννίτες της ΤΝ. Φτάσαμε στο σημείο να τους διορθώνει το Παπαδαριό. www.lifo.gr/now/… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering | Masked diffusion language models (dLLMs) generate text by iteratively denoising masked positions, re-predicting each token multiple ti… | ↗Not briefed |
rss bing-news-ai | 1mo | Corporate data privacy concerns fuel the rise of sovereign artificial intelligence | Data privacy has emerged as a crucial consideration for businesses evaluating artificial intelligence vendors. According to Fortune, c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | anthropic and openai are both dead men walking unless they get some bailout or nationalization, neither of which our economy or national de… | anthropic and openai are both dead men walking unless they get some bailout or nationalization, neither of which our economy or nation… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Since at least the early 2010s, lobbyists from tech companies have been crawling all over govt and there is a revolving door for advisors a… | Since at least the early 2010s, lobbyists from tech companies have been crawling all over govt and there is a revolving door for advis… | ↗Not briefed |
bluesky bluesky-ai | 1mo | does Casey Newton’s partner still work at Anthropic or nah | does Casey Newton’s partner still work at Anthropic or nah | ↗Not briefed |
hn hn-ai | 1mo | Color Psychology in UI Design: How Color Shapes User Behavior | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | According to new reporting from the New York Times, the company has for months been wining and dining religious scholars across the world i… | According to new reporting from the New York Times, the company has for months been wining and dining religious scholars across the wo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Optimization Landscape of Learning Compacted Context Models | Many works approach continual learning through the lens of infinite context windows. As an agent puts more observation into context (c… | ↗Not briefed |
hn hn-ai | 1mo | The reciprocal sum of the prime-prefix-free numbers converges [pdf] | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | There are conferences, and then, there's Goatmire. It was such an incredible experience that I'm struggling to find the words to summarize… | There are conferences, and then, there's Goatmire. It was such an incredible experience that I'm struggling to find the words to summ… | ↗Not briefed |
bluesky bluesky-ai | 1mo | oh yeah the other one i HATE and which is surely going to blow up soon is privacy. ok, most of you all on here hate these tech and would ne… | oh yeah the other one i HATE and which is surely going to blow up soon is privacy. ok, most of you all on here hate these tech and wou… | ↗Not briefed |
hn hn-ai | 1mo | (Lean) Doom | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Mulligan: Performance-Guided Data Collection for Efficient On-Robot Learning | Learning from human demonstrations is a reliable way to teach robots new tasks, but the gains from each additional demonstration shrin… | ↗Not briefed |
hn hn-ai | 1mo | Tmux-agent-status – Quiet, keyboard-first control of coding agents | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning to Learn a Language | We present the Prior-Fitted Language Model (PFLM), a 300M-parameter byte-level transformer pretrained only on samples from a synthetic… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly, nothing is more stressful than running a dry run with small local LLMs for my talk at @devoxx.com . My models are being really mo… | Honestly, nothing is more stressful than running a dry run with small local LLMs for my talk at @devoxx.com . My models are being real… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OGAM: Connecting Systematic Testing to Runtime Assurance through Object-Grounded Attention Monitoring for VLA Policies | Benchmarks expose vision-language-action (VLA) policies to few canonical instructions, while exhaustive deployment testing is impossib… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Mooring – a better screen sharing experience for Macs | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Piratages par des agents d’OpenAI, démission chez Anthropic, crainte d’une IApocalypse… Le chercheur américain Michael I. Jordan revient su… | Piratages par des agents d’OpenAI, démission chez Anthropic, crainte d’une IApocalypse… Le chercheur américain Michael I. Jordan revie… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Not great! | Not great! | ↗Not briefed |
arxiv arxiv-ai | 1mo | Global Communication or Graph-Specific Memory? | Scalable Graph Transformers are commonly trained and evaluated on static large graphs in a transductive setup. Many scalable Graph Tra… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Off-Policy Merging Beats On-Policy Self-Distillation for Continual Learning | A long-standing goal of AI is a model that can continually learn and improve itself. On post-trained models, supervised finetuning (SF… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Finite-Sample Distribution Theory and Efficient Large-Scale Inference for Online Quantile Regression | This paper studies online quantile regression for large-scale and streaming data using Stochastic SubGradient Descent (SSGD) with cons… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Weave Mamba Fusion: Global Cross-Scale Interaction for Lightweight Face Detection | Feature pyramid methods, from FPN to BiFPN, have achieved strong performance in face detection by fusing multi-scale features. However… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CCQ: A Multi-State Child Care Quality Dataset to Support AI for Children's Health Research | High-quality child care in early life is a critical determinant of children's growth and development. Research on child care quality h… | ↗Not briefed |
bluesky bluesky-ai | 1mo | これまでMSやAdobeやその他のベンダが膨大な人月を費やして築いてきたコードの分厚い蓄積が他企業にとっての参入障壁になり、その市場の寡占化を許してきたわけだけど、LLMによってコーディングのコストが激安になった結果、今まで積み重ねてきたコードベース全体の価値が剥落し始めている… | これまでMSやAdobeやその他のベンダが膨大な人月を費やして築いてきたコードの分厚い蓄積が他企業にとっての参入障壁になり、その市場の寡占化を許してきたわけだけど、LLMによってコーディングのコストが激安になった結果、今まで積み重ねてきたコードベース全体の価値が剥落し… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Imagine to Act: High-Fidelity Data Synthesis via Image Editing World Model for Scalable GUI Agent Training | Graphical User Interface (GUI) agents have emerged as a promising paradigm for automating complex digital workflows across diverse app… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Curriculum Brain: Constructing Curriculum Knowledge Graphs as a Substrate for Cognitive Diagnosis | Cognitive Diagnostic Models (CDMs) identify which specific skills a student has and has not mastered, the signal a personalized learni… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI: 28 Days of Codex Improvements – Or a Full Usage Reset | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Overwhelming evidence proves it takes a human to create An OpenAI model was caught trying to cheat at StarCraft, and of course it did it b… | Overwhelming evidence proves it takes a human to create An OpenAI model was caught trying to cheat at StarCraft, and of course it did… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Employé démissionnaire d'OpenAI David Robinson, y a passé trois ans et demi Et dénonce une culture de l’approximation: Un insuffisante cul… | Employé démissionnaire d'OpenAI David Robinson, y a passé trois ans et demi Et dénonce une culture de l’approximation: Un insuffisant… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Rare Failure Eval | — | ↗Not briefed |
hn hn-ai | 1mo | Split an ultrawide monitor into two real displays on macOS | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hierarchical Reinforcement Learning for Collision-Free Locomotion of an Underactuated Biped | A bipedal robot cannot deviate from its path to avoid an obstacle without disturbing its balance, and this coupling is most severe on… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 知人と「そのうちパッケージアプリって簡単にクローンできるようになるかもね」と話してたら、Adobe CCのスケールで始まってた。 逆コンパイル/逆アセンブルみたいな強引な手法に頼らなくても、提供される機能とUIとマニュアルが手に入れば、LLMで商用アプリケーションのクローンが… | 知人と「そのうちパッケージアプリって簡単にクローンできるようになるかもね」と話してたら、Adobe CCのスケールで始まってた。 逆コンパイル/逆アセンブルみたいな強引な手法に頼らなくても、提供される機能とUIとマニュアルが手に入れば、LLMで商用アプリケーションの… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TurboPairFormer: Fast and Stable Protein Folding Model Training with an Optimized Triangle Attention Kernel | Triangular attention is a core computation in AlphaFold3-style biomolecular models, with cubic cost in token count. Its shared pair bi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Blind Spot Paradox: When Adaptive Classifiers Defeat Drift Detectors | Monitoring concept drift from an adaptive classifier's error stream creates an operational conflict with the model's own update loop.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RepICL: Reusable In-Context Prediction Across Heterogeneous Representation Spaces | Frozen representations are widely reused for downstream classification, yet each new task typically requires fitting a new predictor.… | ↗Not briefed |