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
2moDyMT-ESB: Dynamic Multi-Turn Evaluation of Social Bias in User-LLM InteractionsWarning: This paper contains examples of stereotypes and social bias. LLMs are increasingly used in interactive settings by the genera…
↗Not briefed
bluesky
bluesky-ai
2moLa presse française a de nouveau plongé sans recul dans la psychose à cause d'un tweet affirmant que l'IA a "10% de chances" de nous exterm…La presse française a de nouveau plongé sans recul dans la psychose à cause d'un tweet affirmant que l'IA a "10% de chances" de nous e…
↗Not briefed
rss
techmeme
2moSources: OpenAI and Anthropic staff felt blindsided by Dario Amodei's and Sam Altman's calls to slow the frontier; some fear evaluators may compromise security (Cristina Criddle/Financial Times)Cristina Criddle / Financial Times: Sources: OpenAI and Anthropic staff felt blindsided by Dario Amodei's and Sam Altman's calls to sl…
↗Not briefed
arxiv
arxiv-ai
2moFallacy Benchmarks Measure Scheme Recognition, Not Fallacy DetectionFallacy-detection benchmarks pair fallacy classes with a single "valid" or "none" class that takes everything data collection did not…
↗Not briefed
hn
hn-ai
2moDoctors got transplanted testicular tissue in an infertile patient to grow sperm—
↗Not briefed
arxiv
arxiv-ai
2moSTRETCH the Boundaries: A Unified Self-Taught Framework for Progressive LLM EvolutionLarge language models (LLMs) often suffer from capability stagnation in self-improvement training because fixed difficulty levels fail…
↗Not briefed
arxiv
arxiv-ai
2moCoRe-MARL: Cooperative Redistribution Under Unknown Dynamics Using Recurrent Multi-Agent Reinforcement LearningEmergency management assistance programs, such as relief distribution, are essential for delivering necessary supplies to affected com…
↗Not briefed
bluesky
bluesky-ai
2moNB: AI is not going to wipe out humanity. Anthropic are calling for a slowdown because they have no path to profit and are about run out …NB: AI is not going to wipe out humanity. Anthropic are calling for a slowdown because they have no path to profit and are about run…
↗Not briefed
rss
techcrunch-ai
2moSK Hynix reportedly in talks with Intel to build memory chips in USSK Hynix told TechCrunch the company hasn't finalized any plans or arrangements yet.
↗Not briefed
bluesky
bluesky-ai
2moYou’re missing the “our weird sex cult has created robot god and you must advance its interests or it will torture you for all eternity and…You’re missing the “our weird sex cult has created robot god and you must advance its interests or it will torture you for all eternit…
↗Not briefed
arxiv
arxiv-ai
2moGenStream: Semantic Streaming Framework for Generative Reconstruction of Human-centric MediaVideo streaming dominates global internet traffic, yet conventional pipelines remain inefficient for structured, human-centric content…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI investors have approached the company about a new funding roundOpenAI investors have approached the company about a new funding round
↗Not briefed
bluesky
bluesky-ai
2monot to jump on the glorified classifier bandwagon but this model has limited utility and in the modern world will ALREADY be piloted by llm…not to jump on the glorified classifier bandwagon but this model has limited utility and in the modern world will ALREADY be piloted b…
↗Not briefed
hn
hn-ai
2moHackers Got Inside a Flock Camera. Its Data Shows How the System Works—
↗Not briefed
hn
hn-ai
2moWhy building a Rust LSP is hard—
↗Not briefed
rss
techmeme
2moSources: Apple is developing enterprise AI servers powered by two or four M8 Ultra chips, set for 2029, and is considering integrating Nvidia's NVLink Fusion (The Information)The Information: Sources: Apple is developing enterprise AI servers powered by two or four M8 Ultra chips, set for 2029, and is consid…
↗Not briefed
bluesky
bluesky-ai
2moThe AI doomerism isn’t a “hoax”. It’s an egomaniacal Slacktivist cash grab, mixed with media ignorance, political opportunism, and transg…The AI doomerism isn’t a “hoax”. It’s an egomaniacal Slacktivist cash grab, mixed with media ignorance, political opportunism, and t…
↗Not briefed
bluesky
bluesky-ai
2mo"The idea that preparing students to succeed in an AI-saturated world requires reimagining college as an AI incubator is not a sign of tech…"The idea that preparing students to succeed in an AI-saturated world requires reimagining college as an AI incubator is not a sign of…
↗Not briefed
bluesky
bluesky-ai
2moI actually don't disagree with the speaker. We just need to make openAI and Anthropic liable for anything a computer program does on their …I actually don't disagree with the speaker. We just need to make openAI and Anthropic liable for anything a computer program does on t…
↗Not briefed
bluesky
bluesky-ai
2moI already see one dodge that is being built in from the 'oh no we have to slow down!' is that they can claim 'regulation' slowed them down …I already see one dodge that is being built in from the 'oh no we have to slow down!' is that they can claim 'regulation' slowed them…
↗Not briefed
bluesky
bluesky-ai
2moStupid game, forcing tough duel when you know nothing about it, cant progress unless you beat this wally. Thanks not playing your #Aniimo …🔧 ChatGPTStupid game, forcing tough duel when you know nothing about it, cant progress unless you beat this wally. Thanks not playing your #Ani…
↗Not briefed
bluesky
bluesky-ai
2mo"Oh, current harms you say?""Oh, current harms you say?"
↗Not briefed
arxiv
arxiv-ai
2moHow Many Labels Does Model Choice Need? Certificates and Budgets for Selective PredictionClassifiers can make identical predictions yet require labels to compare their selective performance: confidence ranks weight the same…
↗Not briefed
bluesky
bluesky-ai
2moMozilla gaat samenwerken met Mistral, een Europese AI-ontwikkelaar. Mistral Small 4 wordt voortaan een van de modellen voor de ingebouwde A…Mozilla gaat samenwerken met Mistral, een Europese AI-ontwikkelaar. Mistral Small 4 wordt voortaan een van de modellen voor de ingebou…
↗Not briefed
hn
hn-ai
2moShow HN: Kaflow Search – A desktop tool for locally indexed Kafka message searchFor a long time, I’ve used Kafka for services for some purpose. like tracing user action log, event sourcing propagation pipeline step…
↗Not briefed
hn
hn-ai
2moBenchmark Llamacpp on Intel Panther Lake: Vulkan vs. SYCL vs. OpenVINO vs. CPU—
↗Not briefed
bluesky
bluesky-ai
2moHEY DEEPSEEK AI: Tell Me About How China's $22 Laptops Are Destroying A $30 Billion PC Empire And Put In A Video Format That I Can Post On …HEY DEEPSEEK AI: Tell Me About How China's $22 Laptops Are Destroying A $30 Billion PC Empire And Put In A Video Format That I Can Pos…
↗Not briefed
bluesky
bluesky-ai
2moElon Musk on artificial general intelligence: "AGI is significantly higher risk than nuclear weapons, in my opinion. Super smart humans hav…Elon Musk on artificial general intelligence: "AGI is significantly higher risk than nuclear weapons, in my opinion. Super smart human…
↗Not briefed
hn
hn-ai
2moOpenAI Considers Pre-IPO Funding Round at More Than $1.2T Valuation—
↗Not briefed
arxiv
arxiv-ai
2moLearning Array Signal Topologies as Conditional Neural ManifoldsSubspace methods such as multiple signal classification (MUSIC) achieve super-resolution direction of arrival (DoA) estimation by expl…
↗Not briefed
bluesky
bluesky-ai
2moAI psychosis: thinking that newer versions of LLM suddenly operate in an entirely new, magical and fundamentally different way than older m…AI psychosis: thinking that newer versions of LLM suddenly operate in an entirely new, magical and fundamentally different way than ol…
↗Not briefed
arxiv
arxiv-ai
2moWeakening Neurons: An Input-Output Functionality in Transformers with Outsize InfluenceWe analyze the learned input-output behavior of GLU-based neurons in large language models (LLMs). We propose a simple analysis method…
↗Not briefed
hn
hn-ai
2moShow HN: Picxel – Turn reference images into pixel-art game assetsPicxel helps indie game developers redraw references as recognizable game assets. GPT interprets the subject and directs the redraw; l…
↗Not briefed
arxiv
arxiv-ai
2moA Geometric Theory of Decision Boundaries in Structured Markov Decision ProcessesClassical dynamic programming represents optimal sequential decisions through value functions and policies. While this functional repr…
↗Not briefed
hn
hn-ai
2moImpactGate: A merge gate that scores the structural decay AI adds—
↗Not briefed
rss
techcrunch-ai
2moFormer Infosys chief’s AI startup nabs another $53MThe Palo Alto startup says it has landed multiple seven-figure enterprise contracts within months of launch.
↗Not briefed
rss
openai-news
2moReimagining advertising with AIExplore new AI-powered advertising experiences from OpenAI, including Sponsored Agents, tools for marketers, and integrations with Hub…
↗Not briefed
arxiv
arxiv-ai
2moPACT: Can Enterprise AI Assistants Be Trusted Under Pressure?As corporate AI adoption continues to grow, enterprise-grade LLM agents are being deployed into sensitive contexts such as hiring, hea…
↗Not briefed
bluesky
bluesky-ai
2moWe are once again remined that no one in #IlDouche's regime actually reads anything. For the benefit of VP #WeirdIAm I would note that the …We are once again remined that no one in #IlDouche's regime actually reads anything. For the benefit of VP #WeirdIAm I would note that…
↗Not briefed
hn
hn-ai
2moHow Electric Cars Took Hold in Australia—
↗Not briefed
rss
techmeme
2moModular data center startup Rune raised a $40M Series A led by Spark Capital to deploy off-grid compute capacity directly at solar and renewable energy sites (Adele Peters/Fast Company)Adele Peters / Fast Company: Modular data center startup Rune raised a $40M Series A led by Spark Capital to deploy off-grid compute c…
↗Not briefed
hn
hn-ai
2moBrowser Automation Skill from Tencent—
↗Not briefed
arxiv
arxiv-ai
2moOnline Robust Reinforcement Learning Through Monte-Carlo PlanningMonte Carlo Tree Search (MCTS) is a powerful framework for solving complex decision-making problems, yet it often relies on the assump…
↗Not briefed
arxiv
arxiv-ai
2moHypothesis-Driven Autonomous Materials Synthesis with Multimodal LLM AgentsSelf-driving laboratories can explore synthesis conditions autonomously, but their decision-making layer is typically a black-box opti…
↗Not briefed
arxiv
arxiv-ai
2moReasoning through Evolution: Automatic Meta-path Discovery for LLM-based Fake News DetectionPropagation structures provide crucial evidence for fake news detection, yet existing approaches primarily rely on supervised GNN-base…
↗Not briefed
hn
hn-ai
2moUseful Abstractions—
↗Not briefed
arxiv
arxiv-ai
2moReDIL-GNN: Resynthesis Domain Incremental Learning for Circuit Graph Neural NetworksLogic resynthesis preserves circuit functionality while changing gate vocabulary, topology, and structural statistics, creating domain…
↗Not briefed
bluesky
bluesky-ai
2moReports on the OpenAI-Hugging Face incident now stand as an account of harm through which public institutions should examine OpenAI’s decis…🔧 Hugging FaceReports on the OpenAI-Hugging Face incident now stand as an account of harm through which public institutions should examine OpenAI’s…
↗Not briefed
arxiv
arxiv-ai
2moRecursive Reasoning or Statistical Extrapolation? In-Context Learning in Multi-Agent Interdependent Decision-MakingIn-context learning (ICL) enables large language model (LLM) agents to improve decisions using interaction history, yet it remains unc…
↗Not briefed
hn
hn-ai
2moIs METR a Meaningful Check on Anthropic?—
↗Not briefed