AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
arxiv
arxiv-ai
2moLearning Fractional-Order Dynamics from a Single TrajectoryMany real-world processes exhibit long-range dependence, where the current state depends on a slowly decaying trace of past states rat…
↗Not briefed
arxiv
arxiv-ai
2moDesigning Agentic AI Workflow Portfolios under Imperfect Selection and Compute CostAgentic AI systems often approach the same task through multiple workflows that differ in reasoning strategy, verification structure,…
↗Not briefed
bluesky
bluesky-ai
2moI told her Anthropic is basically all guys like this who think OpenAI isn’t aware of the Skynet apocalypse enough and she thought I was bei…I told her Anthropic is basically all guys like this who think OpenAI isn’t aware of the Skynet apocalypse enough and she thought I wa…
↗Not briefed
rss
techmeme
2moSources: investors recently approached OpenAI with a proposal to invest at a $1.2T valuation, but OpenAI believes its value should be at least $1.5T (Sri Muppidi/New York Times)🔧 ChatGPTSri Muppidi / New York Times: Sources: investors recently approached OpenAI with a proposal to invest at a $1.2T valuation, but OpenAI…
↗Not briefed
rss
bing-news-ai
2moMIT, Suffolk Release Joint Study on How Artificial Intelligence is Transforming the Construction IndustrySuffolk, one of the most innovative and successful builders in the country, in partnership with the MIT Center for Real Estate and MIT…
↗Not briefed
rss
bing-news-ai
2moMIT, Suffolk Release Joint Study on How Artificial Intelligence is Transforming the Construction IndustrySuffolk, one of the most innovative and successful builders in the country, in partnership with the MIT Center for Real Estate and MIT…
↗Not briefed
rss
marktechpost
2moNums AI Releases Causilo: A Tabular Foundation Model That Tops TabArena Among Single ModelsNums AI has released Causilo, a pretrained tabular foundation model for classification and regression with a scikit-learn interface. I…
↗Not briefed
bluesky
bluesky-ai
2moI am worried about the asshole humans who are using AI to amplify their shittiness. Missing in all those breathless chatter is the Anthropi…I am worried about the asshole humans who are using AI to amplify their shittiness. Missing in all those breathless chatter is the Ant…
↗Not briefed
hn
hn-ai
2moDatamimic – don't let your coding agent invent its own test world—
↗Not briefed
arxiv
arxiv-ai
2moAutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines🔧 vLLMLarge language model agents tune GPU kernels and serving engines through a closed loop of propose, measure, and keep, but the measurem…
↗Not briefed
arxiv
arxiv-ai
2moPentestChain: A Cost-Aware, MCP-Orchestrated Framework for Automated Penetration Testing with Free-Tier LLMs🔧 Ollama🔧 OpenRouter🔧 CerebrasAI-driven penetration testing has been demonstrated with premium frontier models such as GPT-4, but the per-engagement token cost make…
↗Not briefed
hn
hn-ai
2moAsk HN: How do you put together a technical engineering team?when you are hiring a team for new projects or making changes to it. I expected obvious difficulties, like budget limits, or lots of a…
↗Not briefed
arxiv
arxiv-ai
2moPreservation of Log-Concavity and Convergence of Wasserstein-Fisher-Rao Gradient FlowsWe study the convergence of Wasserstein-Fisher-Rao (WFR) gradient flows for sampling from probability distributions known up to a norm…
↗Not briefed
rss
techmeme
2moTypeSafe AI debuts Jev, a model using "Reinforcement Learning for Calibrated Decisions" to produce typed probabilistic decisions that software can use directly (Thomas Claburn/The Register)Thomas Claburn / The Register: TypeSafe AI debuts Jev, a model using “Reinforcement Learning for Calibrated Decisions” to produce type…
↗Not briefed
bluesky
bluesky-ai
2moCool then point to the clause that says user data and inputs won't be viewed by humans either at Open AI nor humans of any 3rd party contra…Cool then point to the clause that says user data and inputs won't be viewed by humans either at Open AI nor humans of any 3rd party c…
↗Not briefed
bluesky
bluesky-ai
2moGoogle DeepmindのAI研究者が退職し「AIが人類を絶滅させる可能性がある」と警告 https://gigazine.net/news/20260916-google-deepmind-engineer-resigned/Google DeepmindのAI研究者が退職し「AIが人類を絶滅させる可能性がある」と警告 https://gigazine.net/news/20260916-google-deepmind-engineer-resigned/
↗Not briefed
hn
hn-ai
2moSGLang and Miles Add Day-0 Support for DeepSeek-v4.1—
↗Not briefed
hn
hn-ai
2moGLiClass: Open-Source JEV—
↗Not briefed
bluesky
bluesky-ai
2moYeah, woke up thinking this. The #AILaunch is likely to be a politically staged event where some number of companies claim they are so defi…Yeah, woke up thinking this. The #AILaunch is likely to be a politically staged event where some number of companies claim they are so…
↗Not briefed
bluesky
bluesky-ai
2mothis antibench philosophy post is a bit buried, but assuming it's directionally correct, im keeping an open mind. at the model's cost level…this antibench philosophy post is a bit buried, but assuming it's directionally correct, im keeping an open mind. at the model's cost…
↗Not briefed
arxiv
arxiv-ai
2moToken Latency Fairness: Performance Isolation for Multi-Tenant LLM ServingLLM serving is typically offered as a shared, multi-tenant service, where high-demand workloads from one client can cause latency SLO…
↗Not briefed
bluesky
bluesky-ai
2mo👀 per CNBC: WELL OF COURSE CONSODERRING ZUCK'S GOT A BUNKER BUILT IN HAWAII. 🤡 " Meta CEO Mark Zuckerberg sides with Nvidia's Huang o…👀 per CNBC: WELL OF COURSE CONSODERRING ZUCK'S GOT A BUNKER BUILT IN HAWAII. 🤡 " Meta CEO Mark Zuckerberg sides with Nvidia's Hu…
↗Not briefed
bluesky
bluesky-ai
2moEven as Donald Trump blasts Anthropic’s Dario Amodei over his call to hit the brakes on AI development, the two agree on the need to preven…Even as Donald Trump blasts Anthropic’s Dario Amodei over his call to hit the brakes on AI development, the two agree on the need to p…
↗Not briefed
arxiv
arxiv-ai
2moA Comprehensive Review of Generative Physical Artificial Intelligence🔧 ModalThe integration of large-scale foundation models with physical embodiments has led to significant advancements in robotics known as Ge…
↗Not briefed
hn
hn-ai
2moGPT-6 Astra vs. GPT-5.6 Sol: Is a 1.6x Higher Cost Worth It per Verified Bug?🔧 GPT-5—
↗Not briefed
arxiv
arxiv-ai
2moFoundAna: A GNN-assisted Foundation Model for Graph Anomaly DetectionGraph anomaly detection aims to identify graph structures (e.g., nodes, edges, or subgraphs) that deviate significantly from expected…
↗Not briefed
bluesky
bluesky-ai
2moIt’s not just laziness, it’s an LLM! And honestly? okay I don’t hate myself enough to finish this bitIt’s not just laziness, it’s an LLM! And honestly? okay I don’t hate myself enough to finish this bit
↗Not briefed
bluesky
bluesky-ai
2moMy mental model of AI as it exists is that it’s a force multiplier for things people already know how to do. Like the danger isn’t “machine…My mental model of AI as it exists is that it’s a force multiplier for things people already know how to do. Like the danger isn’t “ma…
↗Not briefed
bluesky
bluesky-ai
2moare you using an LLM to write theseare you using an LLM to write these
↗Not briefed
arxiv
arxiv-ai
2moLinguistic Triggers of Gender and Racial Bias in Open-Weight LLMs Applied to Recruitment🔧 Qwen 3🔧 Gemma 3Open-weight large language models are rapidly entering hiring pipelines, yet their discriminatory failure modes -- and the regulatory…
↗Not briefed
bluesky
bluesky-ai
2moah, missed this link in the blog post. there is some more detail on the evals. the tasks are interesting, they're like decision trees. very…ah, missed this link in the blog post. there is some more detail on the evals. the tasks are interesting, they're like decision trees.…
↗Not briefed
bluesky
bluesky-ai
2moThe frontier Anthropic & OpenAI models are only ~4 months ahead of open models — at 5x the price. “Most organizations should ideally be usi…The frontier Anthropic & OpenAI models are only ~4 months ahead of open models — at 5x the price. “Most organizations should ideally b…
↗Not briefed
arxiv
arxiv-ai
2moiMINDBench: iEEG Multi-Institution Neural Decoding BenchmarkIntracranial electroencephalography (iEEG) is widely used to record electrical activity directly from electrodes inside the human brai…
↗Not briefed
arxiv
arxiv-ai
2moBeyond Pixel Similarity: Task-Aware Evaluation of GAN-Based Synthetic Sonar Data for Robotic PerceptionSynthetic data can reduce the cost of collecting and annotating training data for robotic perception, but generating sensor observatio…
↗Not briefed
arxiv
arxiv-ai
2moWhen Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented GenerationGraph-based retrieval-augmented generation (RAG) can help answer questions that require information from many documents. However, buil…
↗Not briefed
bluesky
bluesky-ai
2moThe report identifies seven categories of misuse: cyber operations, surveillance, influence operations, conventional weapons, biological mi…The report identifies seven categories of misuse: cyber operations, surveillance, influence operations, conventional weapons, biologic…
↗Not briefed
hn
hn-ai
2moOpenAI breakthrough triggers 'existential crisis' in math—
↗Not briefed
hn
hn-ai
2moShow HN: 2049 - a roguelike that fuses 2048 and Minesweeper—
↗Not briefed
bluesky
bluesky-ai
2moDear Apple, I'm not going to install #macOS Golden Gate, until you provide a switch disabling everything AI, including preventing the downl…Dear Apple, I'm not going to install #macOS Golden Gate, until you provide a switch disabling everything AI, including preventing the…
↗Not briefed
hn
hn-ai
2moShow HN: Context Freshness Ledger – a tiny schema for reviewable agent memory—
↗Not briefed
arxiv
arxiv-ai
2moAgora: Git as Shared Memory for Collective AutoResearchAutonomous research loops such as AutoResearch show that one coding agent can improve a training setup unattended. Run several of them…
↗Not briefed
bluesky
bluesky-ai
2monah Jensen Huang is a king we can't ban nuclear weapons, and open source LLMs are probably dangerous enough as it is the only AI law we n…nah Jensen Huang is a king we can't ban nuclear weapons, and open source LLMs are probably dangerous enough as it is the only AI law…
↗Not briefed
bluesky
bluesky-ai
2moThis is how Anthropic got its name. The idea being that they're there to prevent human extinction by developing "safe" and "aligned" AI mo…This is how Anthropic got its name. The idea being that they're there to prevent human extinction by developing "safe" and "aligned"…
↗Not briefed
rss
techmeme
2moGoogle researchers help develop the first complete digital map of an adult male fruit fly's brain, which modders programmed to play Doom, trade BTC, and more (Becky Ferreira/404 Media)Becky Ferreira / 404 Media: Google researchers help develop the first complete digital map of an adult male fruit fly's brain, which m…
↗Not briefed
hn
hn-ai
2moShow HN: ZenTabTimer, an HTML timer where browser favicon is the timerAn HTML page where the favicon is a circle that drains over timer. Great for pomodoro work sessions while keeping the timer in view. L…
↗Not briefed
arxiv
arxiv-ai
2moFedPGT: Progressive Gradient Transmission for Vehicular Federated Learning over Time-Varying ChannelsVehicular federated learning (VFL) enables privacy-preserving collaborative model training for intelligent transportation systems, whe…
↗Not briefed
hn
hn-ai
2moInference Is the Last LLM Moat—
↗Not briefed
arxiv
arxiv-ai
2moMask 2D-3D: Adaptive Dual-Masked Autoencoder Network for Image-to-Point Cloud Registration🔧 ModalDetection-free methods for image-to-point cloud registration are prone to erroneous correspondences caused by domain and modality disc…
↗Not briefed
hn
hn-ai
2moWe struggled with DevOps, so built a open source agentic platform just for thisAs a developer with some years of experience and with 150+ projects completed, me and my team learnt devops the hard way. Multiple fix…
↗Not briefed
arxiv
arxiv-ai
2moNot All Layers Need Tuning: Diagnosing and Directing Adaptation in Vision-Language-Action ModelsFine-tuning a Vision-Language-Action (VLA) model for a new deployment environment is expensive, yet most methods apply uniform-capacit…
↗Not briefed