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
bluesky
bluesky-ai
2mo“The labs are saying to governments: ‘We’re quite likely to kill every human being on Earth; please stop us.’ Governments respond with ‘Can…“The labs are saying to governments: ‘We’re quite likely to kill every human being on Earth; please stop us.’ Governments respond with…
↗Not briefed
hn
hn-ai
2moShow HN: I built a browser-native agent for automating job application forms—
↗Not briefed
arxiv
arxiv-ai
2moImproving Reduced-Order Rotating Detonation Engine Models with Data Assimilation and Machine LearningRotating detonation engines (RDEs) exhibit strongly nonlinear, multiscale wave dynamics that set the observed thermal field. High-fide…
↗Not briefed
bluesky
bluesky-ai
2moIkke skriv noe til KIen du ikke vil at en stakkars fyr hos OpenAi må lese på jobbenIkke skriv noe til KIen du ikke vil at en stakkars fyr hos OpenAi må lese på jobben
↗Not briefed
bluesky
bluesky-ai
2moAnthropic’s view of AI safety is basically: 1. Smarter-than-human AI is coming. 2. Major risks are China gets there first or it goes pape…Anthropic’s view of AI safety is basically: 1. Smarter-than-human AI is coming. 2. Major risks are China gets there first or it goes…
↗Not briefed
bluesky
bluesky-ai
2moBorja Adsuara, experto en estrategia digital, sobre el avance de la IA: "Estamos en un momento fascinante de la historia donde todo es posi…Borja Adsuara, experto en estrategia digital, sobre el avance de la IA: "Estamos en un momento fascinante de la historia donde todo es…
↗Not briefed
bluesky
bluesky-ai
2moRiddle me this. WHY would OpenAI suddenly be pushing for this? Why now? Call me a cynic, but this smells suspicious as fuck. And I’m PRO …Riddle me this. WHY would OpenAI suddenly be pushing for this? Why now? Call me a cynic, but this smells suspicious as fuck. And I’m…
↗Not briefed
bluesky
bluesky-ai
2mothe thing about this is, a lot of the steps in the chain show horrible safety practices and embarrassing failures on the part of the evalua…the thing about this is, a lot of the steps in the chain show horrible safety practices and embarrassing failures on the part of the e…
↗Not briefed
bluesky
bluesky-ai
2moI think it’s pretty clear at this point that people will pay for AI (this is not a moral judgement, just a factual statement based on how m…I think it’s pretty clear at this point that people will pay for AI (this is not a moral judgement, just a factual statement based on…
↗Not briefed
rss
techmeme
2moSources: OpenAI bought Glass Imaging, which is developing AI-powered smartphone camera tech, in a deal valuing it at $300M+; it was valued at ~$100M last year (Wall Street Journal)Wall Street Journal: Sources: OpenAI bought Glass Imaging, which is developing AI-powered smartphone camera tech, in a deal valuing it…
↗Not briefed
hn
hn-ai
2moDo agents call 911 when given the option?—
↗Not briefed
arxiv
arxiv-ai
2moSceneBench: A Hierarchical Benchmark for Vision-Language Understanding of 3D ScenesVision-language models excel at 2D image understanding but remain limited in 3D spatial reasoning. Progress is hindered by limitations…
↗Not briefed
arxiv
arxiv-ai
2moToward Governance-Aware Autonomous GIS: A Narrative Review of Ethical and Privacy Risks in LLM-Enabled GeoAIGeospatial artificial intelligence (GeoAI) powered by large language models (LLMs) is expanding the capacity to query, generate, and i…
↗Not briefed
arxiv
arxiv-ai
2moTest-Time Unlearning via Sparse AutoencoderMachine unlearning aims to remove specific knowledge from a trained large language model (LLM) without retraining from scratch. Existi…
↗Not briefed
bluesky
bluesky-ai
2moone thing you gotta keep in mind about the rationalist AI safety people at Anthropic and OpenAI is that a) they genuinely believe what they…one thing you gotta keep in mind about the rationalist AI safety people at Anthropic and OpenAI is that a) they genuinely believe what…
↗Not briefed
hn
hn-ai
2moAnthropic Disrupts Bioweapons Efforts, Russian Hacking, Chinese Claude Misuse—
↗Not briefed
hn
hn-ai
2moShow HN: Serverside.chat—
↗Not briefed
bluesky
bluesky-ai
2moOs modelos chineses pararam de correr atrás de ter o maior número de parâmetros e tão investindo em refinar a tecnologia dentro do limite v…Os modelos chineses pararam de correr atrás de ter o maior número de parâmetros e tão investindo em refinar a tecnologia dentro do lim…
↗Not briefed
rss
techmeme
2moHouse Speaker Johnson says there's "potentially" a role for Congress in creating AI guardrail legislation and that he plans to hold a meeting with AI executives (Erik Wasson/Bloomberg)Erik Wasson / Bloomberg: House Speaker Johnson says there's “potentially” a role for Congress in creating AI guardrail legislation and…
↗Not briefed
arxiv
arxiv-ai
2moHow I learned to stop worrying and love StopGrads: Stationarity, Convergence, and a case study on Flow Map LearningStopgrads are widely used in training machine learning models, but stopgrads can alter the gradient, stationary points and convergence…
↗Not briefed
bluesky
bluesky-ai
2mo“OpenAI acquires smartphone camera startup Glass Imaging” — Investing.com Nigeria #StartupNews #OpenAI #Startups“OpenAI acquires smartphone camera startup Glass Imaging” — Investing.com Nigeria #StartupNews #OpenAI #Startups
↗Not briefed
bluesky
bluesky-ai
2moNow that's a banger of a quote.. "Perhaps at some point governments will remember that their voters prefer not to be dead" www.ft.com/cont…Now that's a banger of a quote.. "Perhaps at some point governments will remember that their voters prefer not to be dead" www.ft.com…
↗Not briefed
bluesky
bluesky-ai
2moEveryone should go and make a monster out of twigs RIGHT NOW with the proviso that you’re not allowed to ask an LLM how to make a monster o…Everyone should go and make a monster out of twigs RIGHT NOW with the proviso that you’re not allowed to ask an LLM how to make a mons…
↗Not briefed
arxiv
arxiv-ai
2moWhere Should the KV Cache Live? Placement Policies Across GPU, CPU, and SSD for Long-Lived SessionsGPU high bandwidth memory is scarce and expensive, and KV caches consume much of it as chats, agent loops, and document question answe…
↗Not briefed
bluesky
bluesky-ai
2moDon't know if I'd ever use a cheap one for work but I tossed DeepSeek v4.1 Flash an idea that's been floating around in my head about takin…Don't know if I'd ever use a cheap one for work but I tossed DeepSeek v4.1 Flash an idea that's been floating around in my head about…
↗Not briefed
bluesky
bluesky-ai
2moThis newsletter has an incredible business model that is unfortunately very misunderstood. www.bloomberg.com/opinion/news...This newsletter has an incredible business model that is unfortunately very misunderstood. www.bloomberg.com/opinion/news...
↗Not briefed
bluesky
bluesky-ai
2moI'm not even sure I really believe we're getting that Anthropic money—no, we are, probably, but also, lemme see that check clear first—but …I'm not even sure I really believe we're getting that Anthropic money—no, we are, probably, but also, lemme see that check clear first…
↗Not briefed
arxiv
arxiv-ai
2moArtificial intelligence and biosecurity: capabilities, threat pathways, and defense-in-depth governanceArtificial intelligence is reshaping biological research across an increasingly connected digital-to-physical workflow. General-purpos…
↗Not briefed
bluesky
bluesky-ai
2moAs empresas de IAs-LLM não conseguem gerar lucro e já estavam com problemas de financiamento. Hoje, o petróleo chega a 110 dólares e, cons…As empresas de IAs-LLM não conseguem gerar lucro e já estavam com problemas de financiamento. Hoje, o petróleo chega a 110 dólares e,…
↗Not briefed
hn
hn-ai
2moOpenSwiftUI – open-source implementation of Apple's SwiftUI—
↗Not briefed
hn
hn-ai
2moJj-stack: stacked GitHub pull requests for Jujutsu—
↗Not briefed
hn
hn-ai
2moI started my SaaS 3 months ago and Indian businesses are absuing the free trialI have a b2b product which solves certain problems for marketers.It&#x27;s called LocalAds<p>We have been growing organically on Insta…
↗Not briefed
arxiv
arxiv-ai
2moCalibrate, Then Route: A Measured Study of Learned Request Routing for Disaggregated LLM Serving🔧 vLLMDisaggregated LLM serving places compute heavy prefill and memory heavy decode on separate GPU pools. Systems such as DistServe, Split…
↗Not briefed
hn
hn-ai
2moShow HN: Typed agent memory with corrections and history—
↗Not briefed
arxiv
arxiv-ai
2moDecoy Direction Optimization: A Post-Hoc Defense Against LLM AbliterationSafety guardrails in open-weight language models can be readily bypassed using Refusal Feature Ablation (RFA), a technique that identi…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic CEO Amodei and his rival at Open AI, Sam Altman agree the industry should “pace the frontier”, or slow growth This is, presumabl…Anthropic CEO Amodei and his rival at Open AI, Sam Altman agree the industry should “pace the frontier”, or slow growth This is, pres…
↗Not briefed
bluesky
bluesky-ai
2moso many things about the rust tooling that can be improved by a determined expert/team member with a mandate and an LLMso many things about the rust tooling that can be improved by a determined expert/team member with a mandate and an LLM
↗Not briefed
bluesky
bluesky-ai
2mowww.thebeaverton.com/2026/09/tech...www.thebeaverton.com/2026/09/tech...
↗Not briefed
bluesky
bluesky-ai
2moI think one lesson from how the LLM thing is going is that intelligence is embodied, and this isn't just a coincidental property of human i…I think one lesson from how the LLM thing is going is that intelligence is embodied, and this isn't just a coincidental property of hu…
↗Not briefed
hn
hn-ai
2moOpenAI Buys Startup Developing Smartphone Camera—
↗Not briefed
arxiv
arxiv-ai
2moCorrupt Plans, Clean Traces: Evading Chain-of-Thought Monitoring with Plan InjectionChain-of-thought (CoT) monitoring is a safety strategy where the reasoning of a large language model "actor" is inspected by a "monito…
↗Not briefed
bluesky
bluesky-ai
2moWIRED featured our study on a phenomenon happening alongside the mainstream AI publishing controversies: Some people are using AI to write …WIRED featured our study on a phenomenon happening alongside the mainstream AI publishing controversies: Some people are using AI to w…
↗Not briefed
arxiv
arxiv-ai
2moA Sentinel-2 benchmark dataset for deep-learning active-fire segmentation across 25 California wildfiresThis article describes an open image dataset for developing and evaluating active-fire segmentation methods in satellite imagery. The…
↗Not briefed
hn
hn-ai
2moImproving Throughput by Optimising KV Cache Efficiency for Agentic Workloads—
↗Not briefed
bluesky
bluesky-ai
2moHeb je al eens informatie aan Chat GPT, de Google-AI of een andere LLM gevraagd over iets waar je echt veel van wist? En klopte het? Mijn e…Heb je al eens informatie aan Chat GPT, de Google-AI of een andere LLM gevraagd over iets waar je echt veel van wist? En klopte het? M…
↗Not briefed
hn
hn-ai
2moEPA eliminates rule that limits planet-warming greenhouse gas emissions from pp—
↗Not briefed
bluesky
bluesky-ai
2moMore reporting on how scientists were trying to create bioweapons and how that was blocked by Anthropic. Again, that is not correct - Anth…More reporting on how scientists were trying to create bioweapons and how that was blocked by Anthropic. Again, that is not correct -…
↗Not briefed
hn
hn-ai
2moBotanical illustrations by women shape our knowledge of Australia's flora—
↗Not briefed
hn
hn-ai
2moRust cargo vet, go vet.. meet PHP vet—
↗Not briefed
hn
hn-ai
2moShow HN: 100 Player Online Smash BrosI used GPT-6 Astra and the recent Melee decomp project (<a href="https:&#x2F;&#x2F;github.com&#x2F;doldecomp&#x2F;melee" rel="nofollow…
↗Not briefed