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
2moThe maths and destruction of humanity stories are distinct to me. A safety researcher at Anthropic said that they think that there's a >10…The maths and destruction of humanity stories are distinct to me. A safety researcher at Anthropic said that they think that there's…
↗Not briefed
bluesky
bluesky-ai
2moシンギュラリティで人類が滅ぶのが現実になりつつあるかもしれない。 Anthropicの研究者が退社し超知能の開発競争を猛批判──「来年末には制御不能になる可能性」 - ITmedia NEWS www.itmedia.co.jp/news/article...シンギュラリティで人類が滅ぶのが現実になりつつあるかもしれない。 Anthropicの研究者が退社し超知能の開発競争を猛批判──「来年末には制御不能になる可能性」 - ITmedia NEWS www.itmedia.co.jp/news/article...
↗Not briefed
bluesky
bluesky-ai
2moOpenAI is begging for someone to slow the AI race #AxiosOpenAI is begging for someone to slow the AI race #Axios
↗Not briefed
bluesky
bluesky-ai
2moAnthropic researcher quits saying AI developers believe 'it could kill us all'Anthropic researcher quits saying AI developers believe 'it could kill us all'
↗Not briefed
arxiv
arxiv-ai
2moMulti-Pass, Multi-View Blended Learning for High-Fidelity Volumetric CT Synthesis from Chest X-RaysReconstructing volumetric Computed Tomography (CT) from a single 2D chest radiograph (CXR) is an ill-posed inverse problem, further co…
↗Not briefed
hn
hn-ai
2moShow HN: Impress your boss with interactive Scikit-Learn Decision Tree—
↗Not briefed
bluesky
bluesky-ai
2moWhenever I hear one of those 'I've quit (AI Model company) because they are building a superintelligence that will wipe out mankind within …Whenever I hear one of those 'I've quit (AI Model company) because they are building a superintelligence that will wipe out mankind wi…
↗Not briefed
bluesky
bluesky-ai
2moAnd it can be destroyed without regret or moral scruple. This will transform policing and warfare in ways that science fiction has grappled…And it can be destroyed without regret or moral scruple. This will transform policing and warfare in ways that science fiction has gra…
↗Not briefed
bluesky
bluesky-ai
2mo会社の偉い人と話してたら趣味の開発は一部ローカルLLMで開発してるとか言われてこいつカッコ良すぎると思った会社の偉い人と話してたら趣味の開発は一部ローカルLLMで開発してるとか言われてこいつカッコ良すぎると思った
↗Not briefed
hn
hn-ai
2moShow HN: Project Estimator – a free, no sign-up software cost estimatorI got tired of pricing projects by gut feeling, so I built this after numerous cost estimation errors that costed me extra hours of fr…
↗Not briefed
arxiv
arxiv-ai
2moDevelopment and Validation of a Physics-Guided Machine Learning Extrapolation Framework Using a Classical Transient Diffusion BenchmarkMachine learning models used in engineering are typically trained within limited operating ranges, yet reliable predictions are often…
↗Not briefed
bluesky
bluesky-ai
2mo"Nos metimos en esto porque la semana pasada circulaban rumores en internet de que los modelos de Anthropic habían resuelto un problema del…"Nos metimos en esto porque la semana pasada circulaban rumores en internet de que los modelos de Anthropic habían resuelto un problem…
↗Not briefed
bluesky
bluesky-ai
2moWell, that's comfortingWell, that's comforting
↗Not briefed
bluesky
bluesky-ai
2moWhy is Anthropic pressing ahead with AI, even though some of its staff think it could wipe out humanity? Wrote about this recently for @th…Why is Anthropic pressing ahead with AI, even though some of its staff think it could wipe out humanity? Wrote about this recently fo…
↗Not briefed
bluesky
bluesky-ai
2moI like how LLM fans like to say "it's not just fancy autocorrect" when like, my dude, I have used so many of these models in so many situat…I like how LLM fans like to say "it's not just fancy autocorrect" when like, my dude, I have used so many of these models in so many s…
↗Not briefed
arxiv
arxiv-ai
2moA Kernel-Based Modular Discriminant Analysis Framework for Small-Sample LearningThe small-sample-size (SSS) problem remains a fundamental challenge in machine learning when labeled data are scarce due to cost, acce…
↗Not briefed
hn
hn-ai
2moSteam now requires Australian users to verify their age with a credit card—
↗Not briefed
arxiv
arxiv-ai
2moMeta-LinEXP3: Online-within-Online Learning for Adversarial Linear Contextual BanditsMeta-learning has emerged as an effective paradigm for transferring knowledge across sequential bandit tasks. While substantial progre…
↗Not briefed
arxiv
arxiv-ai
2moFlowCPO: A Unified Divergence View of Preference Alignment for Flow ModelsPreference alignment for flow and diffusion models now spans online reinforcement learning and offline preference optimization, but th…
↗Not briefed
bluesky
bluesky-ai
2moLa historia detrás del culebrón de la semana sobre la resolución de uno de los problemas del milenio por OpenAI y el posible papel de dos m…La historia detrás del culebrón de la semana sobre la resolución de uno de los problemas del milenio por OpenAI y el posible papel de…
↗Not briefed
arxiv
arxiv-ai
2moBeyond Conventional Federated Learning via High-Order RegularizationFederated clients that perform several local optimization steps can return parameter displacements with widely different magnitudes. T…
↗Not briefed
arxiv
arxiv-ai
2moContrastive Projection: Reading Transformer Internals by Differencing Logit LensesReading a transformer's internal states in token space is easy to do and hard to trust: a logit lens on a single hidden state is domin…
↗Not briefed
bluesky
bluesky-ai
2moHarlan Ellison: In my short story, I Have No Mouth, and I Must Scream, I invented AM as a cautionary tale. Also I died eight years ago. An…Harlan Ellison: In my short story, I Have No Mouth, and I Must Scream, I invented AM as a cautionary tale. Also I died eight years ago…
↗Not briefed
rss
bing-news-ai
2moAmericans use artificial intelligence, but still prefer that humans make consequential decisionsArtificial intelligence (AI) is becoming part of everyday life in the U.S., but Americans are far less comfortable when it shifts from…
↗Not briefed
rss
bing-news-ai
2moUAE, France set out to lead the world in artificial intelligence in space with Altair-Next GenUAE and France lead $1B Altair-Next Gen AI satellite constellation with Marlan Space, Loft Orbital and Mistral, delivering real-time i…
↗Not briefed
bluesky
bluesky-ai
2moJacob Coxon hat genug gesehen. Der 27-jährige Brite, der sich auf das Training neuer KI-Modelle spezialisiert hat, hat seinen Job bei Anthr…Jacob Coxon hat genug gesehen. Der 27-jährige Brite, der sich auf das Training neuer KI-Modelle spezialisiert hat, hat seinen Job bei…
↗Not briefed
bluesky
bluesky-ai
2moOgså kjempeinteressant og nyttig om hva som skjer med teksten når du "bare får litt redigeringshjelp" fra en LLMOgså kjempeinteressant og nyttig om hva som skjer med teksten når du "bare får litt redigeringshjelp" fra en LLM
↗Not briefed
bluesky
bluesky-ai
2moAccording to my memory from last night and the wayback machine, OpenAI updated the Navier-Stokes pdf at Tue, 08 Sep 2026 19:09:35 GMT. The …According to my memory from last night and the wayback machine, OpenAI updated the Navier-Stokes pdf at Tue, 08 Sep 2026 19:09:35 GMT.…
↗Not briefed
bluesky
bluesky-ai
2mo¿Quién ha resuelto el problema del milenio? Una guerra entre IAs amenaza la gloria de dos matemáticos españoles¿Quién ha resuelto el problema del milenio? Una guerra entre IAs amenaza la gloria de dos matemáticos españoles
↗Not briefed
arxiv
arxiv-ai
2moDeep and shallow biases in language modelsLarge language models often repeatedly select the same answer even when many alternatives are plausible. Prior work treats this concen…
↗Not briefed
arxiv
arxiv-ai
2moStrangers to Themselves: What Language Models Say About Themselves Is GenericLanguage models can fluently describe how they would behave: whether they would cave to pushback, misuse a tool, or lie under pressure…
↗Not briefed
rss
techmeme
2moGoogle plans to invest €13B in Finland, its biggest investment in Europe to date, over two years to build three new data centers and expand an existing center (Kirsi Heikel/Bloomberg)Kirsi Heikel / Bloomberg: Google plans to invest €13B in Finland, its biggest investment in Europe to date, over two years to build th…
↗Not briefed
bluesky
bluesky-ai
2moBut LLMs do this not quite random but statistically plausible juxtaposition based on analogy. This explains misaligned citations. Usually (…But LLMs do this not quite random but statistically plausible juxtaposition based on analogy. This explains misaligned citations. Usua…
↗Not briefed
hn
hn-ai
2moThe Fundamental Limits of LLMs at Scale—
↗Not briefed
bluesky
bluesky-ai
2mo🔴🇺🇸FLASH - Jacob #Coxon, chercheur de 27 ans, a annoncé sa démission d’#Anthropic et son retrait de l’industrie de l’IA. Plus tôt cette …🔴🇺🇸FLASH - Jacob #Coxon, chercheur de 27 ans, a annoncé sa démission d’#Anthropic et son retrait de l’industrie de l’IA. Plus tôt c…
↗Not briefed
arxiv
arxiv-ai
2moGrounded Evaluation and Repair for NL-to-PDDL Problem GenerationLarge Language Models (LLMs) have shown promise for translating Natural Language (NL) planning descriptions into PDDL problem instance…
↗Not briefed
bluesky
bluesky-ai
2moOjo a esto que va a traer cola. Como dice Terence Tao, y ¿ahora, qué? Si un rumor puede desplegar miles de agentes con millones de dólares …Ojo a esto que va a traer cola. Como dice Terence Tao, y ¿ahora, qué? Si un rumor puede desplegar miles de agentes con millones de dól…
↗Not briefed
bluesky
bluesky-ai
2moHere are Aristotle's analogical metaphors drawn the way an AI scientist might diagram them. Evening is to dawn as age is to youth -> Empedo…Here are Aristotle's analogical metaphors drawn the way an AI scientist might diagram them. Evening is to dawn as age is to youth -> E…
↗Not briefed
bluesky
bluesky-ai
2moLLMs are inherently unpredictable. The people that built them know that. So when they build an agentic loop based on a fixed task sooner or…LLMs are inherently unpredictable. The people that built them know that. So when they build an agentic loop based on a fixed task soon…
↗Not briefed
bluesky
bluesky-ai
2moImagine if you said that the company you work for had a higher than 1 in 10 chance of ending ALL human life... do you think you might get a…Imagine if you said that the company you work for had a higher than 1 in 10 chance of ending ALL human life... do you think you might…
↗Not briefed
arxiv
arxiv-ai
2moProMeta: Few-shot PROTAC-targeted degradation prediction across E3 ligasesProteolysis-targeting chimeras (PROTACs) have emerged as a transformative therapeutic strategy that selectively degrades historically…
↗Not briefed
bluesky
bluesky-ai
2moThe Claude Mythos 5.1 model was only made available to US vetting organisations in a sign of growing protectionism in the industryThe Claude Mythos 5.1 model was only made available to US vetting organisations in a sign of growing protectionism in the industry
↗Not briefed
arxiv
arxiv-ai
2moLeveraging Fine-grained Error Correction in Korean Speech Recognition for Consultation ServicesAutomatic Speech Recognition (ASR) technology is fundamental to customer service automation and large-scale transcription. However, ev…
↗Not briefed
arxiv
arxiv-ai
2moWhen Does Defendant Statement Matter? A Study of Bias and Persuasion in LLM-Simulated JurorsLLMs have been used to simulate human decision-making in professional settings, yet their behaviors in common-law jury trials remain u…
↗Not briefed
hn
hn-ai
2moGambling with our lives: AI researcher quits Anthropic with warning about safety—
↗Not briefed
arxiv
arxiv-ai
2moDecision Transformer for UAV-Mounted RIS-Assisted Dynamic D2D CommunicationsThis paper studies unmanned aerial vehicle (UAV)-mouted reconfigurable intelligent surface (RIS)-assisted device-to-device (D2D) commu…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic withheld latest AI model from UK testing agency - www.ft.com/content/560e... via @FTAnthropic withheld latest AI model from UK testing agency - www.ft.com/content/560e... via @FT
↗Not briefed
bluesky
bluesky-ai
2moOpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hoursOpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hours
↗Not briefed
arxiv
arxiv-ai
2moAlbedo Estimation via Latent Bridge MatchingRecent advances in Intrinsic Image Decomposition (IID) have increasingly relied on generative models. However, progress remains limite…
↗Not briefed
bluesky
bluesky-ai
2moLo paradójico de el hilo de imágenes de los 80 sin IA es que es posible que acaben formando parte del dataset futuro de un entretenimiento …Lo paradójico de el hilo de imágenes de los 80 sin IA es que es posible que acaben formando parte del dataset futuro de un entretenimi…
↗Not briefed