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 | Efficiently Approximating Attention Is Hard | Softmax attention is ubiquitous in modern machine learning, but its quadratic scaling with sequence length makes it costly. To reduce… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Tuesday (9/29) News Roundup: “Anthropic ‘warns of existential AI risks to humanity’”; ICE’s “quiet but ‘gruesome’ surge”; "Trump administra… | Tuesday (9/29) News Roundup: “Anthropic ‘warns of existential AI risks to humanity’”; ICE’s “quiet but ‘gruesome’ surge”; "Trump admin… | ↗Not briefed |
hn hn-ai | 1mo | Using Protobuf Everywhere | We all know whenever discussion of protobuf comes, it's almost always along with grpc. But its just a data format, that can be us… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Loss-Guided Pretraining Data Selection for Time-Series Foundation Models | Time series foundation models (TSFMs) are pretrained on heterogeneous collections containing billions of observations, yet their train… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Open-Source Decision Engine | Dynamic rule engine with in built ui and beginner friendly sandbox | ↗Not briefed |
bluesky bluesky-ai | 1mo | Selling the AI agents that cause chaos and the enterprise safety suite to monitor them in the same investor deck. Incredible business model… | Selling the AI agents that cause chaos and the enterprise safety suite to monitor them in the same investor deck. Incredible business… | ↗Not briefed |
rss techmeme | 1mo | Sources: Nvidia is in early-stage talks with insurance companies to structure risk-mitigation products to protect lenders against loan defaults by neoclouds (Financial Times) | Financial Times: Sources: Nvidia is in early-stage talks with insurance companies to structure risk-mitigation products to protect len… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Quand une entreprise ambitionne de brûler 518 milliards de dollars en cloud et en calcul (alors qu’elle perd de l’argent), il faut un récit… | Quand une entreprise ambitionne de brûler 518 milliards de dollars en cloud et en calcul (alors qu’elle perd de l’argent), il faut un… | ↗Not briefed |
bluesky bluesky-ai | 1mo | À quelques semaines d’une introduction en bourse record, Anthropic inscrit le risque d’extinction de l’humanité dans la documentation fourn… | À quelques semaines d’une introduction en bourse record, Anthropic inscrit le risque d’extinction de l’humanité dans la documentation… | ↗Not briefed |
hn hn-ai | 1mo | NightKeeper: Self-hosted Laravel Nightwatch Cloud server | — | ↗Not briefed |
hn hn-ai | 1mo | How to Red-Team Your AI Agent's Pull Requests in GitHub Actions | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | V-JEPA Policy: Building Effective World-Action Models on Predictive Visual Latents | World-action models (WAMs) couple future visual-state prediction with action generation. By adapting video generators or image-editing… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic notoriously avoids relying on algorithms and fancy LLM architectures it’s no mistake that they hired Andrej Karpathy, who create… | Anthropic notoriously avoids relying on algorithms and fancy LLM architectures it’s no mistake that they hired Andrej Karpathy, who c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | NanoGPT pretraining runs now take 39.9 seconds for a 124M model(!!) now, 124M is *tiny* so it might not seem relevant. But much of the gai… | NanoGPT pretraining runs now take 39.9 seconds for a 124M model(!!) now, 124M is *tiny* so it might not seem relevant. But much of th… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Quite refreshing seeing LLM sessions _remove_ code! | Quite refreshing seeing LLM sessions _remove_ code! | ↗Not briefed |
arxiv arxiv-ai | 1mo | Codebook-Guided Cross-Modal Knowledge Distillation for Structurally Heterogeneous Features🔧 Modal | Cross-modal knowledge distillation transfers knowledge from a teacher modality to a student modality. Existing feature-level alignment… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Trident: Unifying Guarded Dispatch and Host Execution for PyTorch Triton Workloads | User-written Triton kernels enable high-performance GPU computation within PyTorch, but their end-to-end latency can remain dominated… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Differentiating Bisimulation Metrics: A Framework for Parametric Markov Chain Fitting via Bicausal Optimal Transport | Many problems in sequential decision-making, such as imitation learning from observations, state-space compression, world-model learni… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wichtigster Punkt in der Klage gegen OpenAI: ihrer Software (KI) keine "menschlichen Eigenschaften" verleihen dürfen, sehr gut! KI-Regulie… | Wichtigster Punkt in der Klage gegen OpenAI: ihrer Software (KI) keine "menschlichen Eigenschaften" verleihen dürfen, sehr gut! KI-Re… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Asking for What Was Never Requested: Horizontal and Vertical Proactivity in Agents | An agent that uses tools typically responds to what the user explicitly asks, yet completing the task may require information the user… | ↗Not briefed |
hn hn-ai | 1mo | Splink 5.0.0 Released (probabilistic record linkage at scale) | — | ↗Not briefed |
hn hn-ai | 1mo | What TypeSafe Got Right with the Jev Launch | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I am now a factory worker in the AI feature factory.. leadership just presented the model for how the new AI workflow is and man I’ve never… | I am now a factory worker in the AI feature factory.. leadership just presented the model for how the new AI workflow is and man I’ve… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic ‘warns of existential AI risks to humanity’ in IPO document | Anthropic | The Guardian www.theguardian.com/technology/2... | Anthropic ‘warns of existential AI risks to humanity’ in IPO document | Anthropic | The Guardian www.theguardian.com/technology/2... | ↗Not briefed |
rss techmeme | 1mo | Censorship monitor EngelliWeb says X blocked nearly 500 accounts in Turkey, including those of journalists, economists, and academics, after court orders (Inci Ozbek/Bloomberg) | Inci Ozbek / Bloomberg: Censorship monitor EngelliWeb says X blocked nearly 500 accounts in Turkey, including those of journalists, ec… | ↗Not briefed |
hn hn-ai | 1mo | How to Recover Scammed USDT / Bitcoin / ETH / Cryptocurrency | Finally I have been able to retrieve all my lost funds of about $180k,which I never believed I would get again from an unregulated bro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "all our art is handcrafted" nice "we paid our writers" very good "game features LLM based storytelling system" LOUD BUZZER NOISE | "all our art is handcrafted" nice "we paid our writers" very good "game features LLM based storytelling system" LOUD BUZZER NOISE | ↗Not briefed |
hn hn-ai | 1mo | Show HN: SideKernel – a usable MicroVM sandbox for AI coding agents on macOS | Hey everyone, SideKernel is the result of my capstone project for my MSc at Georgia Tech.<p>There are a lot of sandboxes on the market… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The YOLO S-1: Hey, it might kill us all, but we’ll make a ton of money on the way down… giftarticle.ft.com/giftarticle/... | The YOLO S-1: Hey, it might kill us all, but we’ll make a ton of money on the way down… giftarticle.ft.com/giftarticle/... | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Ismail, a DAW that AI agents operate through text | — | ↗Not briefed |
rss mit-tech-review-ai | 1mo | Making AI an asset, not an expense | When customers talk about AI costs, the conversation usually starts with token prices and ends with access to the latest, most capable… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis | Datalog underpins reasoning tasks such as program analysis, but its programs are hard to write. Existing synthesizers automate this ta… | ↗Not briefed |
hn hn-ai | 1mo | QR: Minimal 0-dep QR code generator and reader | — | ↗Not briefed |
hn hn-ai | 1mo | 'inkjet' tech promises to fix OLED's full-screen brightness limitation [video] | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Seeing Is Not Addressing: Auditing Linguistic Access to Frozen Visual Geometry | Visual distinctions are often finer than those reflected in linguistic conceptualization. Vision-language models exhibit a similar asy… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I keep thinking "Anthropic" ought to be the name of a clothing store I pass by on my way to the movies at my local mall... | I keep thinking "Anthropic" ought to be the name of a clothing store I pass by on my way to the movies at my local mall... | ↗Not briefed |
bluesky bluesky-ai | 1mo | The company said its latest Astra model would undergo more work to meet safety standards, and issued an apology for the way it handled the … | The company said its latest Astra model would undergo more work to meet safety standards, and issued an apology for the way it handled… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Interacting particle guidance for sampling reward-tilted generative priors | Inference-time steering adapts pretrained diffusion and flow-based models to new tasks, e.g., to generate samples from a conditional d… | ↗Not briefed |
rss techmeme | 1mo | Shein reports Q2 revenue up 0.9% YoY to $11B and net income down 66% to $228M, in its first report since a Hong Kong IPO; its stock is ~30% below its IPO price (Tracy Qu/Wall Street Journal) | Tracy Qu / Wall Street Journal: Shein reports Q2 revenue up 0.9% YoY to $11B and net income down 66% to $228M, in its first report sin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Follow the Entities: A Corpus Map for Agentic Search | Answering questions and completing tasks over large document collections often requires connecting evidence spread across multiple doc… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ResComEmb: Effective and Efficient Multimodal Embedding via Residual Homogeneity Compression | Multimodal large language models (MLLMs) have shown strong potential for universal multimodal representation learning. However, existi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CredWise: A Controlled Agentic Decision-Intelligence Framework for Explainable and Auditable Credit-Risk Assessment | Credit-risk prediction is important in banking, but a prediction alone does not explain why an applicant is risky or how it should be… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OptiCom : A Unified Framework for State-Conditioned Composition in LLM-Driven Optimization | Large language models (LLMs) are increasingly deployed to solve complex scientific and practical problems via iterative optimization.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Information Bottleneck-Guided Adaptive Hypergraph Transformer for Brain Disease Diagnosis | Exploring high-order correlations and long-range dependencies in brain networks holds significant value for both neuroscience research… | ↗Not briefed |
rss wired-ai | 1mo | OpenAI Delays Release of Latest Model Over Safety Concerns | The company said its latest Astra model would undergo more work to meet safety standards, and issued an apology for the way it handled… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic ‘warns of existential AI risks to humanity’ in IPO document | Anthropic ‘warns of existential AI risks to humanity’ in IPO document | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Semantic Narrowing: Robust and Efficient LLM Watermarking with Hamming Neighborhoods | Semantic watermarking improves robustness against watermark removal attacks by embedding detectable signals into sentence-level repres… | ↗Not briefed |
bluesky bluesky-ai | 1mo | כותרות - השקת ChatGPT החדש בוטלה: חוקרי OpenAI הזהירו שהמודל לא בטיחותי המודל:🔧 ChatGPT | כותרות - השקת ChatGPT החדש בוטלה: חוקרי OpenAI הזהירו שהמודל לא בטיחותי המודל: | ↗Not briefed |
arxiv arxiv-ai | 1mo | CRJudgeBench: Can AI Detect Plausible but Invalid Code Reviews? | Large language models can generate plausible code-review comments, but such comments may contain technically incorrect claims that mis… | ↗Not briefed |
rss wired-ai | 1mo | Timnit Gebru Believes There Is No ‘Existential Threat’ From AI | One of AI’s fiercest critics believes the doom talk is about founders making money, not saving humanity. | ↗Not briefed |