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
hn
hn-ai
2moDrop in your resume, get a personal website for free—
↗Not briefed
bluesky
bluesky-ai
2moTech oligarchs agree calling AI "too dangerous" way cooler than just admitting it doesn't workTech oligarchs agree calling AI "too dangerous" way cooler than just admitting it doesn't work
↗Not briefed
arxiv
arxiv-ai
2moBackward SDEs-based Diffusion for Physics-Constrained GenerationPretrained score-based diffusion models provide strong unconditional priors, yet enforcing measurement or physics consistency in inver…
↗Not briefed
bluesky
bluesky-ai
2mo"AI rules cannot be written by a small group of commercially aligned companies."A. Gomez, CEO of Cohere 🇨🇦 As Ottawa and Brussels explore…🔧 Cohere"AI rules cannot be written by a small group of commercially aligned companies."A. Gomez, CEO of Cohere 🇨🇦 As Ottawa and Brussels ex…
↗Not briefed
bluesky
bluesky-ai
2moVersion 2.2.74 of the Dialogue System for Unity is live on the Asset Store. articy:draft & Localization updates + AI skills files for agent…Version 2.2.74 of the Dialogue System for Unity is live on the Asset Store. articy:draft & Localization updates + AI skills files for…
↗Not briefed
arxiv
arxiv-ai
2moMore Than Just Access: Generative AI as Communication Intermediary for Blind and Low-Vision Users🔧 ChatGPTGenerative AI (GenAI) tools are increasingly woven into how blind and low-vision (BLV) people communicate, not only with digital infor…
↗Not briefed
arxiv
arxiv-ai
2moNoteVQA: Benchmarking VLMs on Real-Life Questions from Human CommunitiesVision-language models (VLMs) increasingly power consumer-facing AI search, yet evaluating them on the diversity of everyday visual qu…
↗Not briefed
hn
hn-ai
2moHas anyone made a legacy codebase more legible to AI coding agents?Quick Summary: Did you ever optimize a legacy/vibe-coded codebase to improve AI driven development results and avoid fixing a bug…
↗Not briefed
rss
techcrunch-ai
2moOnly at TechCrunch Disrupt 2026: What happens when OpenAI ships your roadmap?If you're building an AI company, the question isn't whether foundation models will continue to evolve. It's whether your company will…
↗Not briefed
bluesky
bluesky-ai
2moEx-funcionário da Anthropic alerta: o risco da IA extinguir a humanidade é real! Confira o comentário de Eduardo Moreira no quadro "Ciência…Ex-funcionário da Anthropic alerta: o risco da IA extinguir a humanidade é real! Confira o comentário de Eduardo Moreira no quadro "Ci…
↗Not briefed
bluesky
bluesky-ai
2mo🚨What do equity punters’ collective histrionics over AI stocks implies for their lenders? In the case of Meta, Oracle and SpaceX, nothing…🚨What do equity punters’ collective histrionics over AI stocks implies for their lenders? In the case of Meta, Oracle and SpaceX, no…
↗Not briefed
bluesky
bluesky-ai
2moAI companies must relinquish some of the power at their disposal ft.trib.al/GmgOvBiAI companies must relinquish some of the power at their disposal ft.trib.al/GmgOvBi
↗Not briefed
arxiv
arxiv-ai
2moEEG-Xplain: Decoding Neural Black-Boxes of EEG Foundation ModelsEEG foundation models such as BIOT, LaBraM, and EEGMamba have achieved remarkable performance in neural signal decoding, but their bla…
↗Not briefed
bluesky
bluesky-ai
2moWhy would the record label pay that ugly bozo millions to look unshowered onstage if they could just use an LLM? The music quality would be…Why would the record label pay that ugly bozo millions to look unshowered onstage if they could just use an LLM? The music quality wou…
↗Not briefed
arxiv
arxiv-ai
2moRESKILL: Explicit Failure Attribution and Structured Repair for Interactive Language AgentsLanguage agents increasingly rely on reusable skills, but post-failure repair is often handled by opaque one-shot reflection: a model…
↗Not briefed
bluesky
bluesky-ai
2moSomeone ask Fuckface how to build a LLM.Someone ask Fuckface how to build a LLM.
↗Not briefed
hn
hn-ai
2moOne Gets Through – Missile Command, but you launch the missiles—
↗Not briefed
bluesky
bluesky-ai
2mogreat, detailed reporting. the focus on is on OpenAI but Anthropic confirmed they also do thisgreat, detailed reporting. the focus on is on OpenAI but Anthropic confirmed they also do this
↗Not briefed
arxiv
arxiv-ai
2moScalability and Performance Evaluation of Federated Learning Frameworks: A Comparative AnalysisThis paper presents a systematic examination and experimental comparison of the prominent Federated Learning (FL) frameworks FedML, Fl…
↗Not briefed
bluesky
bluesky-ai
2moSo this thing is very clear that AI shouldn't resemble people, which makes me think: I don't know that I can evaluate the "helpfulness" of …So this thing is very clear that AI shouldn't resemble people, which makes me think: I don't know that I can evaluate the "helpfulness…
↗Not briefed
arxiv
arxiv-ai
2moProjection-Free Multi-level Algorithms for Stochastic Constrained Compositional OptimizationThis paper studies projection-free algorithms for stochastic constrained multi-level compositional optimization. In this context, the…
↗Not briefed
hn
hn-ai
2moShow HN: Pushie – Get notified with a simple webhookPushie lets you send notifications to yourself from anywhere. Create an account, claim a username, and start pushing to every device y…
↗Not briefed
bluesky
bluesky-ai
2mo"oh yeah that dribble i just copy pasted from my agent? yeah you can't do the same. no no it's just you didn't get the computer that allows…"oh yeah that dribble i just copy pasted from my agent? yeah you can't do the same. no no it's just you didn't get the computer that a…
↗Not briefed
bluesky
bluesky-ai
2moコロナ入ってすぐの頃からセコセコと溜め込み改善し続けてきていた競馬予測AIくんがclaudeによってついに実現可能な結果を出してきていよいよ億万長者への道が(通算5回目くらい)コロナ入ってすぐの頃からセコセコと溜め込み改善し続けてきていた競馬予測AIくんがclaudeによってついに実現可能な結果を出してきていよいよ億万長者への道が(通算5回目くらい)
↗Not briefed
arxiv
arxiv-ai
2moDon't Send What You Don't Need: Question-Guided Token Pruning as a Privacy Defense for Vision-Language Models🔧 ModalVisual Question Answering (VQA) with Vision-Language Models (VLMs) is increasingly used in privacy-sensitive and bandwidth-constrained…
↗Not briefed
bluesky
bluesky-ai
2moIdiocracy.Idiocracy.
↗Not briefed
arxiv
arxiv-ai
2moBenchmarking Intra-Patient 3D Deformable Multimodal Image RegistrationMultimodal image registration is a key component of many clinical workflows, yet it remains challenging because corresponding anatomic…
↗Not briefed
arxiv
arxiv-ai
2moCircuit-MLLM: Topological Logic-Guided Latent-Space Visual Reasoning for Circuit Schematic Understanding🔧 Modal🔧 GPT-5Through pre-training on extensive text and image datasets, current multi-modal large language models (MLLMs) achieve strong performanc…
↗Not briefed
bluesky
bluesky-ai
2moOr Anthropic, et al, or their domestic partners, or the vicious fascist nerds who run shopify and the rest of the tech sector, etc.Or Anthropic, et al, or their domestic partners, or the vicious fascist nerds who run shopify and the rest of the tech sector, etc.
↗Not briefed
bluesky
bluesky-ai
2mowww.theguardian.com/business/202... AI-linked stocks fall after tech bosses call for slowdown in ‘reckless’ developmentwww.theguardian.com/business/202... AI-linked stocks fall after tech bosses call for slowdown in ‘reckless’ development
↗Not briefed
bluesky
bluesky-ai
2moDon’t black box your research: Patterns www.cell.com/patterns/ful... . An editorial about our paper on reproducibility of LLM facilitated r…Don’t black box your research: Patterns www.cell.com/patterns/ful... . An editorial about our paper on reproducibility of LLM facilita…
↗Not briefed
rss
techcrunch-ai
2moSuperhuman acquires YC-backed notetaker Fathom as productivity platforms push for agentic workFathom said that it has over 400,000 monthly active users
↗Not briefed
bluesky
bluesky-ai
2mooomf's post made me want to go on a rant about something I get that many of us are forced to use AI tools in our work (my entire job cente…oomf's post made me want to go on a rant about something I get that many of us are forced to use AI tools in our work (my entire job…
↗Not briefed
arxiv
arxiv-ai
2moCIDERS: Cloud-Edge LLM Collaborative Learning via Accelerating Personalized Bilevel Optimization🔧 ConsensusAmid the rapid advancement of physical-world intelligence, cloud-edge collaborative large language models (LLMs) have emerged as a pro…
↗Not briefed
rss
techmeme
2moFiling: X and SpaceXAI move to dismiss their federal antitrust lawsuit in Texas against Apple, resolving accusations of Apple monopolizing smartphone markets (Mike Scarcella/Reuters)Mike Scarcella / Reuters: Filing: X and SpaceXAI move to dismiss their federal antitrust lawsuit in Texas against Apple, resolving acc…
↗Not briefed
hn
hn-ai
2moA list of 1,325 AI assisted repositories, mined from GitHub—
↗Not briefed
arxiv
arxiv-ai
2moCiteShade: Citation Laundering in Multi-Source Retrieval-Augmented Generation and Its Counterfactual DefenseRetrieval-augmented generation (RAG) grounds a language model's answers on retrieved external knowledge and returns each answer with c…
↗Not briefed
arxiv
arxiv-ai
2moKaininja: Extending Native 3D Generators to the Part LevelNative 3D generators turn one image into a single mesh. TRELLIS.2 and its peers deliver high-fidelity non-watertight geometry with mat…
↗Not briefed
bluesky
bluesky-ai
2moit's pretty funny that "llm chatbot" turned out to be yet another job llm chatbots can't do without close human involvementit's pretty funny that "llm chatbot" turned out to be yet another job llm chatbots can't do without close human involvement
↗Not briefed
bluesky
bluesky-ai
2moReleases later this week! 👀 Photographer: @sergioxangel.bsky.social Creative Director: @torrentempress.neocities.org trueref.ioReleases later this week! 👀 Photographer: @sergioxangel.bsky.social Creative Director: @torrentempress.neocities.org trueref.io
↗Not briefed
arxiv
arxiv-ai
2moPredictive Likelihood Ratios for Language Model Watermark DetectionKeyed watermark detection tests dependence between observed tokens and pseudorandom variables reconstructed from a secret key. Buildin…
↗Not briefed
hn
hn-ai
2moRubyGems Open Source Supply Chain Security and OpenAI—
↗Not briefed
bluesky
bluesky-ai
2moone reason they don't talk about unions is because they can already act collectively when they want to. when Sam Altman got fired, 745 empl…one reason they don't talk about unions is because they can already act collectively when they want to. when Sam Altman got fired, 745…
↗Not briefed
bluesky
bluesky-ai
2moI’m assuming the calls of US LLM companies for an “AI slowdown” means they want regulation to make it illegal to use foreign (Chinese) LLM …I’m assuming the calls of US LLM companies for an “AI slowdown” means they want regulation to make it illegal to use foreign (Chinese)…
↗Not briefed
hn
hn-ai
2moShow HN: What an agent does when anyone can read and rewrite its context—
↗Not briefed
arxiv
arxiv-ai
2moEmpathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMsActivation steering has been used to control traits such as honesty, refusal, and sycophancy, yet supportive empathy is evaluated alon…
↗Not briefed
bluesky
bluesky-ai
2mo"OpenAI is hiring hundreds of contractors who read a massive stream of real users’ ChatGPT prompts, with the prompts sometimes including se…🔧 ChatGPT"OpenAI is hiring hundreds of contractors who read a massive stream of real users’ ChatGPT prompts, with the prompts sometimes includi…
↗Not briefed
hn
hn-ai
2moMore Money Than They Ever Imagined–and No Clue How to Spend It—
↗Not briefed
bluesky
bluesky-ai
2momoderating human <=> llm interactionmoderating human <=> llm interaction
↗Not briefed
rss
techmeme
2moTemporal, which makes open-source software that helps apps, including AI agents, recover from failure, raised $550M led by Lightspeed at a $12.55B valuation (Anhata Rooprai/Reuters)Anhata Rooprai / Reuters: Temporal, which makes open-source software that helps apps, including AI agents, recover from failure, raise…
↗Not briefed