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
2dMoMoVL: Modality-Aware Adaptive Compression for On-Device Sparsely-Activated Vision-Language Models🔧 ModalThe rise of the agentic web has been accompanied by growing demand for on-device Vision-Language Models (VLMs) that perform diverse cr…
↗Not briefed
arxiv
arxiv-ai
2dMEM-STRESS: Longitudinal Field Study of a Wearable-Informed AI Chatbot for Stress ManagementStress is an increasingly prevalent concern, creating a need for accessible and personalized support. Stress-management chatbots prima…
↗Not briefed
bluesky
bluesky-ai
2dA coordinated conspiracy of dipshittery.A coordinated conspiracy of dipshittery.
↗Not briefed
bluesky
bluesky-ai
2dRead today's Australia Briefing for your daily dose of the best of Bloomberg from Down Under and around the worldRead today's Australia Briefing for your daily dose of the best of Bloomberg from Down Under and around the world
↗Not briefed
bluesky
bluesky-ai
2dYou can’t even move an image in Word without an African village teleporting to Siberia. An LLM has no chance.You can’t even move an image in Word without an African village teleporting to Siberia. An LLM has no chance.
↗Not briefed
bluesky
bluesky-ai
2dI don’t understand this Marx reference?I don’t understand this Marx reference?
↗Not briefed
bluesky
bluesky-ai
2dIf we can’t use an LLM to patch COBOL, a language with decades of documentation which runs critical systems but which humans are deeply rel…If we can’t use an LLM to patch COBOL, a language with decades of documentation which runs critical systems but which humans are deepl…
↗Not briefed
hn
hn-ai
2dLibreOffice RCE via ODF Form Datasource JDBC Code Execution—
↗Not briefed
bluesky
bluesky-ai
2d😬 oh, that, yeah....I watched one video this year at the start of the Iran oil aggression, dude was spitting about oil, gold, internationa…😬 oh, that, yeah....I watched one video this year at the start of the Iran oil aggression, dude was spitting about oil, gold, interna…
↗Not briefed
arxiv
arxiv-ai
2dMeasuring the Conservatism of Latent Safety Filters for World ModelsA latent safety filter decides from the latent of a world model whether a planned action is safe. Conformal calibration sets the thres…
↗Not briefed
arxiv
arxiv-ai
2dVariable Length Diffuse Everything: Arbitrary Length Generative Modeling on Arbitrary State SpacesDiffusion and flow-based generative models have achieved remarkable results across diverse applications, but most assume data of fixed…
↗Not briefed
rss
techmeme
2dSources: Wall Street execs and the NSA told Congress this week's outages stemmed from systems overwhelmed by a rush to patch vulnerabilities found by AI models (Jeff Stein/The Washington Sun)Jeff Stein / The Washington Sun: Sources: Wall Street execs and the NSA told Congress this week's outages stemmed from systems overwhe…
↗Not briefed
bluesky
bluesky-ai
2dNo I mean like in the middle of their video they’ll be like “and this video is thanks to my sponsor!” and the sponsor is AnthropicNo I mean like in the middle of their video they’ll be like “and this video is thanks to my sponsor!” and the sponsor is Anthropic
↗Not briefed
arxiv
arxiv-ai
2dStructured Extraction of Diffraction Data from Covalent Organic FrameworksPowder X-ray diffraction (PXRD) is the primary structural evidence for covalent organic frameworks (COFs) and many other materials. Ye…
↗Not briefed
bluesky
bluesky-ai
2dThis is the 5th time this week you’ve claimed RSI openAIThis is the 5th time this week you’ve claimed RSI openAI
↗Not briefed
arxiv
arxiv-ai
2dEvaluating Governed LLM-Driven ML Experimentation: Lifecycle, Provenance, and Failure MetricsAutonomous agents that run ML experiments are usually judged by the best score they reach. That score says nothing about how it was re…
↗Not briefed
arxiv
arxiv-ai
2dSimpler Tiny Recursive ModelsSmall models that execute deep looped computational graphs have shown great experimental results on puzzle-like problems like Sudoku,…
↗Not briefed
arxiv
arxiv-ai
2dFinetuning VLAs for Dexterous Manipulation: The Right Checkpoint Is All You NeedVision-Language-Action models have demonstrated strong manipulation capabilities. A limitation of VLA models is that most manipulation…
↗Not briefed
hn
hn-ai
2dA library that managing and running workloads larger than hardware memoryLibrary Project name: memopro Hi, I'm 4th grade student at university in South Korea. While I experimenting with various AI train…
↗Not briefed
bluesky
bluesky-ai
2dAnthropic IPO: how can I buy shares, and is it worth it?Anthropic IPO: how can I buy shares, and is it worth it?
↗Not briefed
arxiv
arxiv-ai
2dCommunicative On-Policy Distillation: Distilling Black-Box Teacher Capabilities through Communication for Long-Horizon TasksOn-policy distillation (OPD) is a powerful paradigm for improving language-model agents on long-horizon tasks, yet its conventional fo…
↗Not briefed
rss
techmeme
2dTech outages disrupting everything from air travel to banking transactions over the past week have fueled public distrust and paranoia about tech infrastructure (Bloomberg)Bloomberg: Tech outages disrupting everything from air travel to banking transactions over the past week have fueled public distrust a…
↗Not briefed
bluesky
bluesky-ai
2damazing how an LLM does a thing and it's a "man bites dog" storyamazing how an LLM does a thing and it's a "man bites dog" story
↗Not briefed
hn
hn-ai
2dKoka: A Functional Language with Effect Types and Handlers—
↗Not briefed
arxiv
arxiv-ai
2dA Free Lunch: Gradient Averaging Achieves Optimal Statistical Efficiency in Newton SketchWe study statistical inference for online sketched Newton methods, in which each Newton system is approximately solved by an inner ran…
↗Not briefed
hn
hn-ai
2dCénotaphe de Newton (AR on Mobile)—
↗Not briefed
hn
hn-ai
2dViteMCP (From the Creator of FastMCP)—
↗Not briefed
rss
techmeme
2dFAA pauses installations of Juniper routers like the one officials believe contributed to Friday's nationwide ground stop; HPE says its device wasn't defective (Neil Vigdor/New York Times)Neil Vigdor / New York Times: FAA pauses installations of Juniper routers like the one officials believe contributed to Friday's natio…
↗Not briefed
bluesky
bluesky-ai
2dIt's not even just the GPUs, Jensen Huang is responsible for providing resources and money for all of the GenAI startups. He's responsible …It's not even just the GPUs, Jensen Huang is responsible for providing resources and money for all of the GenAI startups. He's respons…
↗Not briefed
bluesky
bluesky-ai
2dsay what you will about sam kriss, he can writesay what you will about sam kriss, he can write
↗Not briefed
arxiv
arxiv-ai
2dCratesBump: A Version-Bump Dataset for the Rust EcosystemWe release CratesBump, the first public dataset for predicting version bumps in the Rust ecosystem, together with semver-extract, the…
↗Not briefed
arxiv
arxiv-ai
2dLodestone: Automated Semantic Discovery of Datacenter Hardware SpecializationHardware specialization can improve datacenter efficiency. Unfortunately, identifying computations that justify dedicated hardware is…
↗Not briefed
bluesky
bluesky-ai
2dthe news: oregon DA suing openai for helping kids develop eating disorders and commit suicide, not having sufficient age verification my m…the news: oregon DA suing openai for helping kids develop eating disorders and commit suicide, not having sufficient age verification…
↗Not briefed
hn
hn-ai
2dUnparalleled Misalignments—
↗Not briefed
bluesky
bluesky-ai
2dOn the cusp of what may be the biggest initial public offering in history, AI giant Anthropic is shooting for the pricing moon: Evening Bri…On the cusp of what may be the biggest initial public offering in history, AI giant Anthropic is shooting for the pricing moon: Evenin…
↗Not briefed
arxiv
arxiv-ai
2dSame Model, Different People: Numerical Execution Contracts and the Reproducibility of Fairness Audits🔧 vLLMFairness audits treat a trained model as a fixed decision function, but the arithmetic that evaluates it is part of that function. Thi…
↗Not briefed
bluesky
bluesky-ai
2dI truly believe that the average tech founder/CEO is at least an order of magnitude less important for the success of their company than li…I truly believe that the average tech founder/CEO is at least an order of magnitude less important for the success of their company th…
↗Not briefed
bluesky
bluesky-ai
2d"Amazon and Google will together own nearly 30% of Anthropic after its IPO" techcrunch.com/2026/11/17/a..."Amazon and Google will together own nearly 30% of Anthropic after its IPO" techcrunch.com/2026/11/17/a...
↗Not briefed
hn
hn-ai
2dSelf-Adapting Language Models—
↗Not briefed
bluesky
bluesky-ai
2dAgain, Epic said "make *anything* you want"... but many prompters were so lazy that they just copy-pasted the theme of the jam and didn't p…Again, Epic said "make *anything* you want"... but many prompters were so lazy that they just copy-pasted the theme of the jam and did…
↗Not briefed
rss
techcrunch-ai
2dNvidia promises that revenue will keep soaring — for nowDespite the turmoil Nvidia is in after the departure of its disgraced founder and CEO Jensen Huang, the company projected that its str…
↗Not briefed
bluesky
bluesky-ai
2dWhether openai or anthropic or any other frontier lab is the proximate cause, the failure lies in a lack of governance. The greed behind th…Whether openai or anthropic or any other frontier lab is the proximate cause, the failure lies in a lack of governance. The greed behi…
↗Not briefed
arxiv
arxiv-ai
2dSafe Personal RSI: Evaluating and Repairing Personal-Agent Updates🔧 GPT-5Kicker is a personal health assistant. Changes to an assistant's reusable instructions can improve task handling while changing what i…
↗Not briefed
arxiv
arxiv-ai
2dToaster: Nonlinear Memory from Cascaded Linear Memories and Target PropagationRecurrent memories let language models process long sequences with a fixed-size state. Linear memories such as DeltaNet and KDA store…
↗Not briefed
bluesky
bluesky-ai
2dalso inspired by @bcantrill.bsky.social's recent talk and @sunshowers.io's recent posting and advocating for higher standards for code I d…also inspired by @bcantrill.bsky.social's recent talk and @sunshowers.io's recent posting and advocating for higher standards for code…
↗Not briefed
bluesky
bluesky-ai
2daje.news/ho7egnaje.news/ho7egn
↗Not briefed
bluesky
bluesky-ai
2dJD Vance is basically an LLM trained on Peter Thielology and Donald Trump's strategic racism.JD Vance is basically an LLM trained on Peter Thielology and Donald Trump's strategic racism.
↗Not briefed
arxiv
arxiv-ai
2dSearching the Pareto Front: Multi-Objective Retrieval for Heterogeneous Scientific ModalitiesIn cosmology, surveys span different sky regions and wavelength ranges, resulting in uneven modality coverage across observational dat…
↗Not briefed
arxiv
arxiv-ai
2dWhen Do LLMs Model the Right Target? Problem Formulation in High-Stakes Risk Prediction🔧 ElicitGenerative AI is increasingly used to assist with data science tasks, yet fundamental questions remain about its ability to correctly…
↗Not briefed
hn
hn-ai
2dAn open letter to OpenAI and Anthropic on cybersecurity access—
↗Not briefed