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 | Almost half of China's workforce is now in the gig economy | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Sam Altman to Decoded: ‘The world should accept some bad things happening’ for the benefits of AI" Go fuck yourself! www.politico.com/ne… | "Sam Altman to Decoded: ‘The world should accept some bad things happening’ for the benefits of AI" Go fuck yourself! www.politico.c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reading, Not Manipulating: Leveraging Router Logits for Multimodal Safety in MoE Vision-Language Models | Vision-language models (VLMs) face compositional safety risks where harmful intent emerges from the interaction between visual and tex… | ↗Not briefed |
bluesky bluesky-ai | 1mo | El Pentágono deja de utilizar las herramientas de IA de Anthropic tras incluir a la empresa en una lista negra, según pudo saber la BBC - B…🔧 ChatGPT | El Pentágono deja de utilizar las herramientas de IA de Anthropic tras incluir a la empresa en una lista negra, según pudo saber la BB… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TRACE: Rollout-Guided Quantization-Aware Training for FP4 Reinforcement Learning of MoE Language Models | Reinforcement learning (RL) for post-training large language models (LLMs) incurs substantial computation and memory overhead during r… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OTel: Open Telco AI Datasets, Benchmarks, and Models | We present Open Telco (OTel), an open telecom AI resource that releases derived telecom datasets for retrieval, reranking, instruction… | ↗Not briefed |
arxiv arxiv-ai | 1mo | No Transformer Beats Six Covariates: Long-Horizon Prediction of Depressive Symptoms from Childhood Essays | Natural language processing (NLP) models can detect depression-related language in text written near the time symptoms are measured, b… | ↗Not briefed |
rss techmeme | 1mo | Sources: Seagate and Toshiba are bidding for TDK's HDD magnetic head business in a multibillion-dollar deal, as they compete for AI data center storage demand (Bloomberg) | Bloomberg: Sources: Seagate and Toshiba are bidding for TDK's HDD magnetic head business in a multibillion-dollar deal, as they compet… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ST-Bench: A Spatial-Temporal Benchmark for Multi-Agent System Generation on Scientific Research Tasks🔧 GPT-5 | The rapid progress of LLM-based multi-agent systems (MAS) has shown that they largely outperform single agents on coding, math, and QA… | ↗Not briefed |
hn hn-ai | 1mo | DOS: Determining System Uptime | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Contrastive Learning for Aspect Representation towards Explainable Recommendation | In this work, we propose a novel recommendation model, CLARER (Contrastive Learning for Aspect Representation towards Explainable Reco… | ↗Not briefed |
bluesky bluesky-ai | 1mo | transcript of my spouse and me talking about Claude | transcript of my spouse and me talking about Claude | ↗Not briefed |
arxiv arxiv-ai | 1mo | Later Is Better: Token Reduction for ViTs Under Distribution Shift🔧 Tome | Training-free token reduction accelerates vision transformers by removing redundant tokens across layers, recovering most of the origi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Trustworthy Method Comparison with AI Judges: Estimation and Design under Order, Batch, and Aggregation Effects | Large language models (LLMs) are increasingly used as judges for automated AI evaluation. A common practice is to randomize prompt seq… | ↗Not briefed |
bluesky bluesky-ai | 1mo | that is a crazy fucking decision. that is a completely unsafe technology unless they are vastly more confident in their ability to read act… | that is a crazy fucking decision. that is a completely unsafe technology unless they are vastly more confident in their ability to rea… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Adversarially Trained Linear Transformers Are Optimal Robust In-Context Learners for Gaussian Mixtures | Adversarial training is one of the most reliable defenses against adversarial attacks, but its high computational cost must generally… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Evidence to Action: How Tool-Using Agents Fail | Tool-using agents make consequential changes to external state, yet correct outcomes do not guarantee that their actions were supporte… | ↗Not briefed |
arxiv arxiv-ai | 1mo | How Well Do LLMs Reason with Noisy Evidence? An Active Visual Reasoning Benchmark | Real-world reasoning rarely reduces to static question answering: agents must actively gather information from tools and sensors that… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: DemoForge – Product demo videos generated from Playwright scripts | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Leo que OpenAI admite que sus modelos se inventan datos y los presentan como reales No me digas 😱. Les ha debido llevar años darse cuenta… | Leo que OpenAI admite que sus modelos se inventan datos y los presentan como reales No me digas 😱. Les ha debido llevar años darse c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Student journalists are doing the work again. And despite the faculty resolution and unanimous student resolution, the chancellor is moving… | Student journalists are doing the work again. And despite the faculty resolution and unanimous student resolution, the chancellor is m… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "We built an artificial baby model that learns language directly from raw sound and trained it to imitate whale speech. Then we looked insi… | "We built an artificial baby model that learns language directly from raw sound and trained it to imitate whale speech. Then we looked… | ↗Not briefed |
bluesky bluesky-ai | 1mo | an Anthropic employee discovered the algorithm and handed it off to them to figure out wtf it is | an Anthropic employee discovered the algorithm and handed it off to them to figure out wtf it is | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cleave: Scaling Tensor Program Optimization via Decoupled Algebraic Search and Operator Scheduling | Optimized kernels such as FlashAttention and FlashDecoding are crucial for accelerating today's large models. Most of them are handwri… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT of course did not such thing. OpenAI and Sam Altman did. This will continue until people stop pretending the machine did it. When l…🔧 ChatGPT | ChatGPT of course did not such thing. OpenAI and Sam Altman did. This will continue until people stop pretending the machine did it. W… | ↗Not briefed |
arxiv arxiv-ai | 1mo | High-dimensional online calibration from harmonic weights | We study the online calibration of multidimensional forecasts over an arbitrary convex set $Y\subseteq\mathbb{R}^d$ relative to an arb… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cite What You Explore: Budget-Aware LLM Reasoning over Medical KGs with Verifiable Evidence | Post-discharge risk prediction from electronic health records (EHRs) is difficult because many dependencies that link discharge-time o… | ↗Not briefed |
bluesky bluesky-ai | 1mo | incredible discourse happening in the s&box server right now | incredible discourse happening in the s&box server right now | ↗Not briefed |
arxiv arxiv-ai | 1mo | Nash Social Welfare for Multi Armed Bandits: Trajectory-wise Expected and High Probability Regret | We study fair multi-armed bandits under the Nash Social Welfare (NSW) objective, which measures performance via the geometric mean of… | ↗Not briefed |
hn hn-ai | 1mo | GTA5 in the Browser | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | i once again stress that every time you blow claude tokens on some shit what you are actually doing is contributing to anthropic's "see, pe… | i once again stress that every time you blow claude tokens on some shit what you are actually doing is contributing to anthropic's "se… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My opinions on LLM and machine learning in general, i.e. tools that have been branded as "AI," are also actually quite nuanced, but my curr… | My opinions on LLM and machine learning in general, i.e. tools that have been branded as "AI," are also actually quite nuanced, but my… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning to Retrieve via Reinforcement Learning in Embedding Space | Dense retrieval models are typically trained with contrastive objectives that learn effective representations but do not directly opti… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Insurance claims to test Altman and Amodei liability for ‘rogue’ AI ft.trib.al/XafxCSL | Insurance claims to test Altman and Amodei liability for ‘rogue’ AI ft.trib.al/XafxCSL | ↗Not briefed |
arxiv arxiv-ai | 1mo | SanSi: A Looped Typed Decision Model for System 1.5 Thinking | Typed decision models answer a declared question without generating text: a decision head returns a probability for each of the declar… | ↗Not briefed |
bluesky bluesky-ai | 1mo | この流れが進むとローカルLLMは反社御用達みたいな扱いになって御禁制AIになるのかなと思ったらけど、そもそもこの勢いで反社が金を集められるのなら反社用データセンターみたいのが発展途上国に普通に建ってダーククラウドLLMみたいのが普通に走り出すのかもしれない。ほんまもんのダークウ… | この流れが進むとローカルLLMは反社御用達みたいな扱いになって御禁制AIになるのかなと思ったらけど、そもそもこの勢いで反社が金を集められるのなら反社用データセンターみたいのが発展途上国に普通に建ってダーククラウドLLMみたいのが普通に走り出すのかもしれない。ほんまもん… | ↗Not briefed |
hn hn-ai | 1mo | QBasic Nibbles in C# | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Saken med OpenAI som hacker seg inn på offisielle steder i Australia minner meg litt om hvordan Russland stadig "mister" droner inn over Na… | Saken med OpenAI som hacker seg inn på offisielle steder i Australia minner meg litt om hvordan Russland stadig "mister" droner inn ov… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PERSIST: Who-What-When Memory Across Sessions for Full-Duplex Spoken Dialogue | Modern voice assistants may be shared by multiple users and should be able to answer questions about earlier conversations such as "Wh… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI fügt generiertem Text Wasserzeichen hinzu, um sie als KI-generiert erkennbar zu machen. Aber nicht überall. #Künstliche Intelligenz | OpenAI fügt generiertem Text Wasserzeichen hinzu, um sie als KI-generiert erkennbar zu machen. Aber nicht überall. #Künstliche Intelli… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Model Plants the Trigger: Answer-Side Backdoor Attacks in Multi-Turn Large Language Models | Safety alignment in Large Language Models (LLMs) remains vulnerable to backdoor attacks. Existing LLM backdoors are almost all input-c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods | Activation steering provides a lightweight and flexible way to control large language model (LLM) behavior. However, effective steerin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Exact Calibration and Sharp Risk Geometry for Volume-Sampled Ridge Regression | We study ridge regression from exactly $s$ distinct rows of a fixed design. Responses are fixed, and only the subset is random. The de… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RefRoute: Decoupling Conditioning Cost from References via Compact Residual Conditioning and Spatial Routing🔧 FLUX | Multi-reference image generation requires preserving the appearance of multiple subjects while composing them into a coherent scene. H… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Stability of Measure-to-Measure Transformers on Sub-Gaussian Data | Transformers have exhibited impressive empirical success across various domains, but their theoretical foundations remain less develop… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A Documentation Wayback Machine with Plakar | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Readout Stability in Prefill-Only Decision Models:Zero-Label Prediction and Inference-Time Compute Allocation | Prefill-only decision models inspired by the Jev model score every candidate in a menu during a single forward pass and never decode,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Old Facts Return: Re-Reads, Reverts, and the Limits of Temporal Memory | A memory system can retire an obsolete value and later restore it merely because the same old statement appears again. A re-read of an… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neuromotor Hierarchy Network: Physiological Inductive Biases for Robust Generalization in sEMG Decoding | Surface electromyography (sEMG) provides a wearable, noninvasive interface to neuromuscular activity for movement decoding and human-c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Mathematical Invariant-Enabled Topological Neural Networks for Molecular and Materials Property Prediction | Existing molecular and materials learning approaches often rely on a limited set of structural representations, which may capture only… | ↗Not briefed |