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
2moCan We Do Interpretable NLI with Graphs Based on Atomic Propositions?While Large Language Model (LLM)-based Natural Language Inference (NLI) systems achieve high accuracy, their decision-making processes…
↗Not briefed
hn
hn-ai
2moI built a Grok Bot inspired theme for Hermes Desktop client—
↗Not briefed
bluesky
bluesky-ai
2moit remains fascinating that google, regularly purveyor of the world's best AI, insists on putting its worst model in front of exactly the p…it remains fascinating that google, regularly purveyor of the world's best AI, insists on putting its worst model in front of exactly…
↗Not briefed
hn
hn-ai
2moCrofAI gets exposed as an OpenRouter wrapper, shuts down🔧 OpenRouter—
↗Not briefed
arxiv
arxiv-ai
2moGeometry of learning dynamics: Gradient descent versus natural gradient on the ridge of optimizationHigh-capacity associative memories based on Kernel Logistic Regression (KLR) exhibit a "Ridge of Optimization" characterized by extrem…
↗Not briefed
arxiv
arxiv-ai
2moSOTER: A Generative Time-Series Foundation Model for Wearable Human Physiological SignalsTime-series foundation models have demonstrated strong cross-domain transfer, yet their common architectural assumptions remain poorly…
↗Not briefed
rss
techmeme
2moA deep dive into on-device and data center inference for robots, including a primer on robot models, deployments, supply chains, the "network wall", and more (SemiAnalysis)SemiAnalysis: A deep dive into on-device and data center inference for robots, including a primer on robot models, deployments, supply…
↗Not briefed
bluesky
bluesky-ai
2moUK must heed warnings from AI experts, minister saysUK must heed warnings from AI experts, minister says
↗Not briefed
hn
hn-ai
2moWhich is the better data analyst? Benchmarking ChatGPT vs. Claude🔧 ChatGPT—
↗Not briefed
arxiv
arxiv-ai
2moOn the disintegration of the stochastic majority vote: From PAC-Bayesian bounds to a self-bounding algorithmWeighted majority votes are central to many successful ensemble methods. PAC-Bayesian theory provides tight generalization guarantees…
↗Not briefed
hn
hn-ai
2moOrthoRoute: GPU Accelerated Autorouting for KiCad—
↗Not briefed
arxiv
arxiv-ai
2moSmarter by the Moment: Environment-Driven Dynamic Policies for Continual LLM ImprovementLarge Language Models (LLMs) have achieved remarkable progress across diverse domains, but continual adaptation to evolving tasks and…
↗Not briefed
bluesky
bluesky-ai
2moDas Tool ist im Wesentlichen eine Trace-/Observability-Funktion. Mit viel Aufwand kann man damit nachvollziehen, welche Schritte ein Agent …Das Tool ist im Wesentlichen eine Trace-/Observability-Funktion. Mit viel Aufwand kann man damit nachvollziehen, welche Schritte ein A…
↗Not briefed
arxiv
arxiv-ai
2moTime-warping estimation via stationarity-based learning of the de-warped signalTime-warping estimation is a fundamental problem in signal processing with applications in bioacoustics, radar, and biomedical analysi…
↗Not briefed
arxiv
arxiv-ai
2moLayers, Sinks, and Scaling: Adaptive Evidence Selection for Multimodal Large Language ModelsMultimodal large language models (MLLMs) can answer knowledge-intensive visual questions by combining visual evidence from images with…
↗Not briefed
bluesky
bluesky-ai
2moI think, given even Sam Altman and Darius Amodei have spent the weekend saying AI safety culture is a mess, this should be relatively uncon…I think, given even Sam Altman and Darius Amodei have spent the weekend saying AI safety culture is a mess, this should be relatively…
↗Not briefed
arxiv
arxiv-ai
2moAvailable but Unclaimed: An Empirical Study of Human-AI Synergy🔧 GPT-5People increasingly reason with large language models (LLMs), yet complementary capabilities do not guarantee outperforming both compo…
↗Not briefed
bluesky
bluesky-ai
2moWe have the frontier hacking models or whatever, with all the attendant publicity from the accidental-on-purpose "escapes", but that's abou…We have the frontier hacking models or whatever, with all the attendant publicity from the accidental-on-purpose "escapes", but that's…
↗Not briefed
bluesky
bluesky-ai
2moim sure ai has value for about 500 scientists, worldwide, that are trying to solve famine & disease... they only need 1, maybe 2 data cente…im sure ai has value for about 500 scientists, worldwide, that are trying to solve famine & disease... they only need 1, maybe 2 data…
↗Not briefed
arxiv
arxiv-ai
2moNoise2Noise Revisited: Training Pair Distributions Dominate Loss Choice in Self-Supervised DenoisingNoise2Noise (N2N) trains denoisers on pairs of independently corrupted observations, eliminating clean references. We stress-test two…
↗Not briefed
rss
techmeme
2moNetherlands-based inference chip startup Euclyd raised a €200M+ Series A co-led by Samsung, Somerset Capital, Scaleup Europe Fund, and Innovation Industries (Kai Nicol-Schwarz/CNBC)Kai Nicol-Schwarz / CNBC: Netherlands-based inference chip startup Euclyd raised a €200M+ Series A co-led by Samsung, Somerset Capital…
↗Not briefed
hn
hn-ai
2moGoogle Artemis – Missing attribution to mobile-use and its contributors—
↗Not briefed
hn
hn-ai
2moClaude Fable 5.1 and GPT 6 Astra controlled my Robot Arm. Who painted better?—
↗Not briefed
hn
hn-ai
2moLeslie Lamport: Some Data on the Frequency of Errors in Mathematics Papers [pdf]—
↗Not briefed
arxiv
arxiv-ai
2moIntegrating the Analytic Hierarchy Process with Large Language Models for Transparent Multi-Criteria Decision-MakingLLMs are increasingly employed in a wide range of decision-making tasks. However, the opacity of their internal reasoning makes it dif…
↗Not briefed
bluesky
bluesky-ai
2mofascinating translated essay seemingly written by a deepseek engineer - first half is him contemplating his own replacement by AI; conclusi…fascinating translated essay seemingly written by a deepseek engineer - first half is him contemplating his own replacement by AI; con…
↗Not briefed
hn
hn-ai
2moBrig: Run coding agents in a MicroVM sandbox—
↗Not briefed
arxiv
arxiv-ai
2moBenchmarking Factual Robustness of LLMs via Multi-conversation PersuasionAs Large Language Models (LLMs) increasingly serve as primary knowledge retrieval interfaces, their robustness against \textit{persuas…
↗Not briefed
bluesky
bluesky-ai
2moQwen Cloud has a $6 token plan, they serve Deepseek V4.1 Flash as well. Not as generous as Opencode Go since they charge 10% for cache hits…Qwen Cloud has a $6 token plan, they serve Deepseek V4.1 Flash as well. Not as generous as Opencode Go since they charge 10% for cache…
↗Not briefed
bluesky
bluesky-ai
2mo- Trump wijst regulering van AI af en noemt waarschuwingen over de veiligheid een “hoax”. Hij zegt dat alleen een “sterke en slimme” presid…- Trump wijst regulering van AI af en noemt waarschuwingen over de veiligheid een “hoax”. Hij zegt dat alleen een “sterke en slimme” p…
↗Not briefed
bluesky
bluesky-ai
2moGod, they really don’t watch science fiction do they? www.bbc.co.uk/news/article...God, they really don’t watch science fiction do they? www.bbc.co.uk/news/article...
↗Not briefed
bluesky
bluesky-ai
2moif anthropic is "ant", openai is therefore "ope"if anthropic is "ant", openai is therefore "ope"
↗Not briefed
hn
hn-ai
2moGoogle DeepMind Staffer Says AI May 'Kill Us All' in Exit Post—
↗Not briefed
bluesky
bluesky-ai
2moyeah, i mean, "AI companies massively overestimating/exaggerating what their technology can do" has been a pattern since they entered the p…yeah, i mean, "AI companies massively overestimating/exaggerating what their technology can do" has been a pattern since they entered…
↗Not briefed
bluesky
bluesky-ai
2moSi la RATP et la SNCF ne me jouent pas des tours, je suis à Lille aujourd'hui pour une conf technique qui se lance, CyberBrew. J'ai choisi …Si la RATP et la SNCF ne me jouent pas des tours, je suis à Lille aujourd'hui pour une conf technique qui se lance, CyberBrew. J'ai ch…
↗Not briefed
bluesky
bluesky-ai
2moSo weird that we can be sure Trump did not, in fact, use AI for this. His comms might be the last place safe from AI. They're unfakeably wo…So weird that we can be sure Trump did not, in fact, use AI for this. His comms might be the last place safe from AI. They're unfakeab…
↗Not briefed
bluesky
bluesky-ai
2moIn a post published on his public WeChat account, DeepSeek engineer Liu Shengyu said he expects AI to soon surpass him at the work he loves…In a post published on his public WeChat account, DeepSeek engineer Liu Shengyu said he expects AI to soon surpass him at the work he…
↗Not briefed
arxiv
arxiv-ai
2moCoverage-Aware Virtual IMU Augmentation for Low-Resource Human Activity RecognitionIMU-based human activity recognition (HAR) enables continuous, privacy-friendly monitoring of daily activities using wearable sensors.…
↗Not briefed
bluesky
bluesky-ai
2mo#AI #Anthropic #Executive #extinction #Fox News #Regulation #Trump #SHUT UP, PIGGY 🐷 🌐 www.newsmart.ch#AI #Anthropic #Executive #extinction #Fox News #Regulation #Trump #SHUT UP, PIGGY 🐷 🌐 www.newsmart.ch
↗Not briefed
bluesky
bluesky-ai
2moAnd this one. makingsciencepublic.com/2026/09/11/f...And this one. makingsciencepublic.com/2026/09/11/f...
↗Not briefed
hn
hn-ai
2moShow HN: Yet another argument parser for Zig—
↗Not briefed
bluesky
bluesky-ai
2moHow exactly is AI going to "kill everyone"? I've been seeing such claims from (mis-)Anthropic former workers but not substantiated in any w…How exactly is AI going to "kill everyone"? I've been seeing such claims from (mis-)Anthropic former workers but not substantiated in…
↗Not briefed
arxiv
arxiv-ai
2moTurn-level Multiscale Density Ratio Estimation for LLM AgentsWith the rapid development of Large language model (LLM), agent systems enhanced by LLMs show huge potential in being able to deal wit…
↗Not briefed
bluesky
bluesky-ai
2moI don’t think many people in England’s schools know that LLM chats are not confidential and never have been. I wonder if that would change…I don’t think many people in England’s schools know that LLM chats are not confidential and never have been. I wonder if that would c…
↗Not briefed
bluesky
bluesky-ai
2mo¿Se puede 'frenar' la IA? Por qué el resto del mundo desconfía del plan de Silicon Valley¿Se puede 'frenar' la IA? Por qué el resto del mundo desconfía del plan de Silicon Valley
↗Not briefed
bluesky
bluesky-ai
2moTop AI companies Anthropic and OpenAI should make products and pace their own progress if necessary without government intervention, David …Top AI companies Anthropic and OpenAI should make products and pace their own progress if necessary without government intervention, D…
↗Not briefed
arxiv
arxiv-ai
2moTAME: Token Attribution and Masking for Emergent misalignmentFine-tuning an aligned language model on narrow, flawed data can induce harmful behavior far outside the training domain, known as eme…
↗Not briefed
arxiv
arxiv-ai
2moBeyond Episodic AI: Cognitive Field Networks for Biologically Inspired Persistent CognitionCognitive Field Theory (CFT) proposes that cognition arises from memory-dressed collective dynamics that generate a persistent macrosc…
↗Not briefed
hn
hn-ai
2moCSS-Tricks in Limbo—
↗Not briefed
arxiv
arxiv-ai
2moConstant Swap Regret in General-Sum Games via Optimistic Transition MatricesWe give deterministic and uncoupled learning dynamics for finite multiplayer general-sum games under full-information feedback that ac…
↗Not briefed