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
12dThe Optimal Rate Constant of Heavy-Ball Momentum on Smooth Strongly Convex FunctionsThe heavy-ball method with a constant step size and a constant momentum parameter accelerates gradient descent on quadratics, but no c…
↗Not briefed
hn
hn-ai
12dPrivate LLM copies can agree on a secret key over a public channel—
↗Not briefed
bluesky
bluesky-ai
12d„Ram Altman“: Bash-Skript misst freien RAM und erstellt tägliche E-Mails www.pcgameshardware.de/RAM-Hardware... #Linux #LinuxNews #LinuxDE…„Ram Altman“: Bash-Skript misst freien RAM und erstellt tägliche E-Mails www.pcgameshardware.de/RAM-Hardware... #Linux #LinuxNews #Li…
↗Not briefed
bluesky
bluesky-ai
12dThis is bad in every way www.ft.com/content/41bf...This is bad in every way www.ft.com/content/41bf...
↗Not briefed
arxiv
arxiv-ai
12dPACE-ECG: Patient-Aware Contrastive ECG Pretraining; Leveraging Longitudinal Structure for Transferable RepresentationsFoundation models for physiological signals are commonly evaluated on in-domain interpretation tasks, but their ability to transfer to…
↗Not briefed
arxiv
arxiv-ai
12dHumanSightBench: A Benchmark for Human Visual Perception in MLLMsMultimodal large language models (MLLMs) have demonstrated strong performance on knowledge-intensive and reasoning-based vision-langua…
↗Not briefed
bluesky
bluesky-ai
12dAnthropic $ANT down 33% from its high two months ago on HyperliquidAnthropic $ANT down 33% from its high two months ago on Hyperliquid
↗Not briefed
hn
hn-ai
12dConsider the Agent: Pity, persuasion and selection in an emerging AI ecology—
↗Not briefed
rss
techmeme
12dAlphabet's Wing plans to invest up to $125M over two years to build a 15,000-square-foot facility in Fort Worth, Texas, to produce tens of thousands of drones (Mark R. Long/Wall Street Journal)Mark R. Long / Wall Street Journal: Alphabet's Wing plans to invest up to $125M over two years to build a 15,000-square-foot facility…
↗Not briefed
arxiv
arxiv-ai
12dReconfigurable Action Theories: A Formal Foundation for Dynamically Composable Agent HarnessesAction theories typically fix their vocabulary and causal laws, whereas agent harnesses can change their tools, providers, and policie…
↗Not briefed
arxiv
arxiv-ai
12dPhysics-Informed Sinusoidal Representation Networks for Damage Identification in Ultrasonic Guided Wave Structural Health MonitoringUltrasonic guided wave (UGW) inspection has become an effective method in structural health monitoring (SHM). While machine learning h…
↗Not briefed
hn
hn-ai
12dShow HN: I run 19 production APIs on a free-tier VPS—
↗Not briefed
arxiv
arxiv-ai
12dRecoGEM: A Serving System for Production-Scale Generative RecommendationGenerative recommendation (GR) is moving industrial recommender systems (RecSys) from scoring candidate items to autoregressively gene…
↗Not briefed
arxiv
arxiv-ai
12dBe My Eye: Learning to See Beyond a Robot's Own PerspectiveRobots operating in the real world are inherently subject to partial observability, particularly in multi-robot collaboration, where i…
↗Not briefed
hn
hn-ai
12dShow HN: Harness Resource Bench – CPU, memory and disk of coding agentsHi HN. I built this to see what coding agents cost the machine they run on, not how well they code.<p>Disclosure: I made smelt, which…
↗Not briefed
hn
hn-ai
12dShow HN: Slopie, Spotify wrapped for your Claude Code and Codex historywanted to just know which model i swear at the most, which model do i thank the most, which model costs me the most. - try it using &#…
↗Not briefed
arxiv
arxiv-ai
12dDepth as a Candidate Source: Portfolio Allocation for Recurrent Test-Time Scaling in Large Language ModelsRecurrent latent reasoning in large lanugage models (LLMs) scales test-time computation by increasing recurrent depth, naturally frami…
↗Not briefed
arxiv
arxiv-ai
12dRepresentational Alignment between Deep Neural Networks and Human Brain in Speech Processing under Audiovisual NoiseSpeech comprehension in the human brain is an incremental process that begins with acoustic processing and advances to linguistic proc…
↗Not briefed
bluesky
bluesky-ai
12dDo humans still need to learn math in the age of AI? Yes — and it may be more important than everDo humans still need to learn math in the age of AI? Yes — and it may be more important than ever
↗Not briefed
rss
techmeme
12dMeta says it is globally expanding Instagram's AI age estimation tech to apply Teen Account protections to teens aged 13 to 17 who misrepresented their age (Jess Weatherbed/The Verge)Jess Weatherbed / The Verge: Meta says it is globally expanding Instagram's AI age estimation tech to apply Teen Account protections t…
↗Not briefed
arxiv
arxiv-ai
12dLearning Evidence-Grounded Reasoning for Context-Aware Time Series ForecastingLLM-driven time series forecasting has become increasingly promising by leveraging contextual features to generate future values direc…
↗Not briefed
hn
hn-ai
12dHow SkyPilot on Slurm and Kubernetes put 1k idle GPU hours a week to work—
↗Not briefed
bluesky
bluesky-ai
12deu não suporto como todo mundo aceita qualquer llm tranquilo sem nenhum problemaeu não suporto como todo mundo aceita qualquer llm tranquilo sem nenhum problema
↗Not briefed
hn
hn-ai
12dWhen PGD fails, is your model safe? I tested it vs. Marabou SMT on traffic signs—
↗Not briefed
rss
techmeme
12dA profile of Alice Huang, a suspected Chinese spy who founded Megaspeed in 2023 and spent years in Jensen Huang's orbit; Megaspeed's Nvidia imports top $4.6B (Mark Anderson/Bloomberg)Mark Anderson / Bloomberg: A profile of Alice Huang, a suspected Chinese spy who founded Megaspeed in 2023 and spent years in Jensen H…
↗Not briefed
bluesky
bluesky-ai
12dこの手の連中が次に言い出すのが「すべての端末向けに配信しないメーカーの責任」「新しい富の再配分の在り方だ」かな。 でも(この場合Gemini4なので)「生成AIにトークン費用を突っ込める財力」とか(ローカルLLMを使うにしても)「自環境に生成環境を組み込める知識」が必要だからむ…この手の連中が次に言い出すのが「すべての端末向けに配信しないメーカーの責任」「新しい富の再配分の在り方だ」かな。 でも(この場合Gemini4なので)「生成AIにトークン費用を突っ込める財力」とか(ローカルLLMを使うにしても)「自環境に生成環境を組み込める知識」が必…
↗Not briefed
arxiv
arxiv-ai
12dAgent4Crypto: A Dual-Reflective and Market-Feedback-Aligned LLM-Based Multi-Agent System for Cryptocurrency TradingCryptocurrency markets provide a natural testbed for autonomous AI agents because they are digital-native, high-frequency, and machine…
↗Not briefed
arxiv
arxiv-ai
12dOdds-Structured Distributional Regression for Interpretable Time-to-Event PredictionTime-to-event prediction is fundamentally challenged by partial observability and aleatoric uncertainty. In clinical medicine, for ins…
↗Not briefed
hn
hn-ai
12dShow HN: MyBox – Isolate your project folder in a Linux boxHi HN,<p>I’ve built MyBox to isolate my development projects in Linux containers.<p>Installing npm modules on my machine didn’t feel r…
↗Not briefed
bluesky
bluesky-ai
12dKét dolgot kell most figyelni: Anthropic részvényértékesítésének előrehaladása, majd november 19-én az ING hitelaukciója (eladnak 500millió…Két dolgot kell most figyelni: Anthropic részvényértékesítésének előrehaladása, majd november 19-én az ING hitelaukciója (eladnak 500m…
↗Not briefed
rss
techmeme
12dUS voters in Harwood, North Dakota rejected a ballot measure that opposed Applied Digital's $3B data center, set to open in 2027, bucking a nationwide backlash (Miriam Waldvogel/Washington Post)Miriam Waldvogel / Washington Post: US voters in Harwood, North Dakota rejected a ballot measure that opposed Applied Digital's $3B da…
↗Not briefed
arxiv
arxiv-ai
12dHierarchical Multi-Channel Tip-Over Prediction via Isolated Temporal Models for Planetary Rover SafetyReliable tip-over prevention is critical for planetary rovers operating in unstructured environments where communication delays preven…
↗Not briefed
hn
hn-ai
12dShow HN: NekoTrail – Open-source Conductor alternative for 16 coding agents—
↗Not briefed
hn
hn-ai
12dCoinbase Gets Final Regulatory Stamp for Trust-Bank Charter—
↗Not briefed
arxiv
arxiv-ai
12dRAIL: Representation-Aligned Imitation Learning for Student-Compatible Teacher PoliciesReinforcement learning (RL) from raw sensory inputs such as images or onboard sensors can be challenging due to high-dimensional obser…
↗Not briefed
rss
techmeme
12dThe US OCC grants Coinbase full approval for a trust bank charter; sources say it could expand Coinbase's stablecoin business and let it issue stablecoins (Olga Kharif/Bloomberg)Olga Kharif / Bloomberg: The US OCC grants Coinbase full approval for a trust bank charter; sources say it could expand Coinbase's sta…
↗Not briefed
rss
the-decoder
12dNvidia espionage scandal weakens Huang's lobby against tighter AI chip export controlsNvidia CEO Jensen Huang has admitted to an extramarital affair with a woman US intelligence agencies classify as a suspected Chinese i…
↗Not briefed
bluesky
bluesky-ai
12dsomehow this story lands a little differently 2 months later 🧐 www.theguardian.com/technology/2...somehow this story lands a little differently 2 months later 🧐 www.theguardian.com/technology/2...
↗Not briefed
hn
hn-ai
12dHours of YouTube and 85 handtyped pages of notes on LLM Inference – Start here—
↗Not briefed
arxiv
arxiv-ai
12dA method to automatically discover symbolic local learning rulesWe present a general method for discovering symbolic local learning rules that can replace backpropagation, applicable, in principle,…
↗Not briefed
bluesky
bluesky-ai
12dReject Ai generated product images, embrace the lost art of poorly editing your product onto royalty-free model photos like the gods intend…Reject Ai generated product images, embrace the lost art of poorly editing your product onto royalty-free model photos like the gods i…
↗Not briefed
arxiv
arxiv-ai
12dCADMA: Capacity-Aware Coverage Decomposition for Generative Model AssessmentEvaluating generative models requires measuring not only sample fidelity but also how well a generated set covers the real data distri…
↗Not briefed
arxiv
arxiv-ai
12dControlling Temporal Pseudo-Label Marginals for Stable Online Test-Time AdaptationContinual Test-Time Adaptation (CTTA) methods update models online using a stream of unlabeled data to overcome distribution shifts en…
↗Not briefed
arxiv
arxiv-ai
12dPrune, Don't Merge: Expert Compression of Mixture-of-Experts LLMs on Multilingual Production TasksMixture-of-Experts language models are cheap to run per token but expensive to hold in GPU memory, which drives the cost of serving th…
↗Not briefed
arxiv
arxiv-ai
12dECOS: Environment-Calibrated On-Policy Self-Distillation for Language AgentsOn-policy self-distillation lets a language agent act as its own Teacher by conditioning on privileged context such as skills written…
↗Not briefed
rss
techmeme
12dAmsterdam police arrested between 150 and 200 protesters for trespassing and violence at a Microsoft data center construction site owned by Pure Data Centers (Jowi Morales/Tom's Hardware)Jowi Morales / Tom's Hardware: Amsterdam police arrested between 150 and 200 protesters for trespassing and violence at a Microsoft da…
↗Not briefed
bluesky
bluesky-ai
12dBeen trying Blender MCP with Claude Code. A few hours later I have a nice looking space craft. No concept images, just a 2 hour back and fo…Been trying Blender MCP with Claude Code. A few hours later I have a nice looking space craft. No concept images, just a 2 hour back a…
↗Not briefed
hn
hn-ai
12dShow HN: TimeScroll – an open-source, local replacement for Rewind on Mac🔧 OllamaHi HN, I&#x27;ve been building TimeScroll since September 2025. It&#x27;s a Mac app that records your screen in the background, so you…
↗Not briefed
arxiv
arxiv-ai
12dA Cost That Practice Accuracy Hides: Worked Solutions in Self-Generated Training Data and the Compositional Use of New RulesA worked solution can make self-generated training targets shorter and cheaper, but accuracy on the practiced operation may hide how w…
↗Not briefed
arxiv
arxiv-ai
12dPARIS: A multi-agent LLM framework for end-to-end, PRISMA-guided systematic literature reviews with pre-synthesis corpus quality assessmentSystematic literature reviews (SRs) remain slow and labor-intensive, and most automation tools address single stages of the workflow.…
↗Not briefed