News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
bluesky bluesky-ai | 18d | mas é foda ver que dois jornalistas experientes de fato não entendem como um LLM funciona a manipulação dos resultados de IA é uma realida… | mas é foda ver que dois jornalistas experientes de fato não entendem como um LLM funciona a manipulação dos resultados de IA é uma re… | ↗Not briefed |
arxiv arxiv-ai | 18d | MMedAgent-Plus: Scaling Multimodal Medical Agents across Modalities, Tasks and Workflows | Medical workflows often require multiple interdependent analyses, but existing specialist and generalist AI models are not designed to… | ↗Not briefed |
arxiv arxiv-ai | 18d | Neighborhood-Aware Unlearning for Generative Recommendation | Generative recommendation predicts items by autoregressively generating semantic identifiers (SIDs), whose tokens are shared across se… | ↗Not briefed |
rss techmeme | 18d | Ramp says it applied to the US OCC for a trust bank charter that, if approved, would allow it to offer fiduciary investment management and digital asset custody (Paige Smith/Bloomberg) | Paige Smith / Bloomberg: Ramp says it applied to the US OCC for a trust bank charter that, if approved, would allow it to offer fiduci… | ↗Not briefed |
bluesky bluesky-ai | 18d | DHS plans to ban nonmilitary federal agencies from using technology developed by DeepSeek, a Chinese AI company. https://www.washingtonsun.… | DHS plans to ban nonmilitary federal agencies from using technology developed by DeepSeek, a Chinese AI company. https://www.washingto… | ↗Not briefed |
arxiv arxiv-ai | 18d | Magnolia: A Persistent-Memory Agent Framework for Computational Chemistry | Chemistry has recently entered a high-data regime as molecular databases from high-throughput platforms or computational studies have… | ↗Not briefed |
hn hn-ai | 18d | Clean up what your apps and agents left running | — | ↗Not briefed |
bluesky bluesky-ai | 18d | Communities Take Charge: Data Center Model Policy Guide #DataCenters #AI webuildprogress.org/data-center-... | Communities Take Charge: Data Center Model Policy Guide #DataCenters #AI webuildprogress.org/data-center-... | ↗Not briefed |
bluesky bluesky-ai | 18d | der smarte techbro mit der schnieken lederjacke könnte doch nicht auf eine nachfrage konkret antworten sondern würde nur weiter wolkig rumb… | der smarte techbro mit der schnieken lederjacke könnte doch nicht auf eine nachfrage konkret antworten sondern würde nur weiter wolkig… | ↗Not briefed |
hn hn-ai | 18d | Show HN: Open-source app that lets LLMs control any iPhone app (no Mac required) | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Which Agent's Skill Should Change? Group-Relative Credit Assignment for Evolving Multi-Agent Teams | Cooperative language-model agents can improve without weight updates by revising reusable skills, but only if a shared team reward is… | ↗Not briefed |
rss techmeme | 18d | Sources: Nscale is in talks for a second deal with Anthropic worth ~$45B to deliver 460MW of computing power at its West Virginia campus using Vera Rubin chips (Bloomberg) | Bloomberg: Sources: Nscale is in talks for a second deal with Anthropic worth ~$45B to deliver 460MW of computing power at its West Vi… | ↗Not briefed |
bluesky bluesky-ai | 18d | A judge ordered Abdulrahman Alfozan to remain in custody, likening the Saudi government’s offer to secure the former OpenAI engineer’s rele… | A judge ordered Abdulrahman Alfozan to remain in custody, likening the Saudi government’s offer to secure the former OpenAI engineer’s… | ↗Not briefed |
arxiv arxiv-ai | 18d | Sparse RL Relaxation for LLM Reasoning | Recent work has explored continuous representations in place of discrete tokens during Chain-of-Thought (CoT) reasoning, often referre… | ↗Not briefed |
bluesky bluesky-ai | 18d | Isn’t this also the basis of the AI religion Anthropic practices or is that a separate thing | Isn’t this also the basis of the AI religion Anthropic practices or is that a separate thing | ↗Not briefed |
rss wired-ai | 18d | OpenAI Engineer Accused of Spying on Executives Ordered to Remain in Custody | A judge ordered Abdulrahman Alfozan to remain in custody, likening the Saudi government’s offer to secure the former OpenAI engineer’s… | ↗Not briefed |
hn hn-ai | 18d | I made a tiny 2 MB screenshot tool for Windows (open source) | — | ↗Not briefed |
hn hn-ai | 18d | Ex-OpenAI employee accused of planting device has Saudi backing: Prosecutor | — | ↗Not briefed |
bluesky bluesky-ai | 18d | look im not even strictly anti-llm but come on, at least show some respect for the craft rather than vibing your way to a new shiny toy you… | look im not even strictly anti-llm but come on, at least show some respect for the craft rather than vibing your way to a new shiny to… | ↗Not briefed |
arxiv arxiv-ai | 18d | Deep catastrophic-risk-aware reinforcement learning | Estimating extreme risk in reinforcement learning becomes increasingly difficult as the confidence level increases, since only a small… | ↗Not briefed |
hn hn-ai | 18d | Star Wars: A Russian Military Satellite Gets Within 200m of One Used by Ukraine | — | ↗Not briefed |
arxiv arxiv-ai | 18d | Profile-DP: Differential Privacy for User-Controlled Profile Release in Personalized LLM Inference | Personalized large language models (LLMs) increasingly rely on profile-conditioned inference, pairing public requests with rich, user-… | ↗Not briefed |
hn hn-ai | 18d | What is still hard to become good at as AI makes coding easier? | Since LLMs came, how I build software changed for so many reasons. Since then, I’ve spent most of my time deciding what I want to buil… | ↗Not briefed |
bluesky bluesky-ai | 18d | Poll: Americans now view Big Tech money and CEOs as Republican Vote these oligarch enablers out! #GOTV apple.news/A75RdWNaISHm... | Poll: Americans now view Big Tech money and CEOs as Republican Vote these oligarch enablers out! #GOTV apple.news/A75RdWNaISHm... | ↗Not briefed |
arxiv arxiv-ai | 18d | A Theoretical Analysis of Environment-Aware Out-of-Distribution Detection for Vision-Language Models🔧 Modal | Continual vision-language models (VLMs) encounter a stream of environments and must also reject inputs that do not belong to the envir… | ↗Not briefed |
arxiv arxiv-ai | 18d | You Are the Trigger: Speaker-Activated Backdoors🔧 Whisper | Speech models are increasingly used as an interface to AI systems, from transcription tools to voice agents. Existing audio backdoors… | ↗Not briefed |
arxiv arxiv-ai | 18d | Language Guidance for Supervised Vision Training: An Empirical Study of Generalization | Deep neural networks have achieved remarkable success on vision benchmarks, yet they continue to struggle with many generalization cha… | ↗Not briefed |
bluesky bluesky-ai | 18d | ... milliers d'euros en compensation. Dommage, j'aurai bien aimé faire saigner un peu Anthropic. | ... milliers d'euros en compensation. Dommage, j'aurai bien aimé faire saigner un peu Anthropic. | ↗Not briefed |
arxiv arxiv-ai | 18d | Diffusion Score Estimation with Shallow Neural Networks: Regularity, Approximation, and Finite-Sample Rates | We study the sample complexity of estimating Ornstein--Uhlenbeck diffusion scores in $\mathbb R^d$ using denoising score matching (DSM… | ↗Not briefed |
arxiv arxiv-ai | 18d | LoWAgent: A Low-Resource Multilingual Benchmark for Web Agents | Web agents are typically evaluated on English-language interfaces, leaving their ability to operate across the world's multilingual we… | ↗Not briefed |
bluesky bluesky-ai | 18d | also, there's a non-zero chance this whole thing was a paid ad from anthropic. | also, there's a non-zero chance this whole thing was a paid ad from anthropic. | ↗Not briefed |
rss arstechnica-tech | 18d | Recently fixed Plex Media vulnerabilities are being exploited to install malware | A researcher says attacks chaining 4 vulnerabilities is hacking unpatched boxes. | ↗Not briefed |
bluesky bluesky-ai | 18d | Anthropic eyeing $45B in compute from Nscale. Nvidia-backed, and an IPO boost if it lands. $45B for compute—who really wins here, Anthropic… | Anthropic eyeing $45B in compute from Nscale. Nvidia-backed, and an IPO boost if it lands. $45B for compute—who really wins here, Anth… | ↗Not briefed |
hn hn-ai | 18d | AnkiGen – in-browser generation of Anki decks between any two languages | — | ↗Not briefed |
arxiv arxiv-ai | 18d | LSA-PAST: Layer-Specific Adaptation and Prototype Transport for Compositional Incremental Learning | Class-Incremental Learning (CIL) enables models to recognize new classes over time while retaining prior knowledge, but many real-worl… | ↗Not briefed |
arxiv arxiv-ai | 18d | The Culture Within: Towards Mechanistic Understanding of Cultural Reasoning in LLMs | Large Language Models (LLMs) increasingly operate across culturally diverse settings, yet existing evaluations largely measure cultura… | ↗Not briefed |
arxiv arxiv-ai | 18d | Preference-Aware Pareto Steering for Multi-Task Speech LLMs | Speech large language models (LLMs) perform multiple tasks, including automatic speech recognition (ASR), timestamp prediction (TS), a… | ↗Not briefed |
bluesky bluesky-ai | 18d | The study of mathematics must advance one careful proof after an OpenAI dump after another careful proof. The study of research writing mus… | The study of mathematics must advance one careful proof after an OpenAI dump after another careful proof. The study of research writin… | ↗Not briefed |
hn hn-ai | 18d | Show HN: Jev in action – task classifier and auto model router in Keen Code | Hi HN!<p>In Keen Code, a coding agent solo-developed by me, I have implemented a task-classifier using Jev and a simple auto-router ba… | ↗Not briefed |
rss bing-news-ai | 18d | MSc artificial intelligence for engineering | The University of Sheffield, UK, is inviting applications for its MSc Artificial Intelligence for Engineering course starting in Septe… | ↗Not briefed |
rss bing-news-ai | 18d | MSc artificial intelligence for engineering | The University of Sheffield, UK, is inviting applications for its MSc Artificial Intelligence for Engineering course starting in Septe… | ↗Not briefed |
rss bing-news-ai | 18d | MSc artificial intelligence for engineering | The University of Sheffield, UK, is inviting applications for its MSc Artificial Intelligence for Engineering course starting in Septe… | ↗Not briefed |
bluesky bluesky-ai | 18d | D'aucuns suggèreraient que ptêt c'est pas un hasard si la plate-forme "permettant" à Anthropic de larguer la grosse thunasse ne fonctionne … | D'aucuns suggèreraient que ptêt c'est pas un hasard si la plate-forme "permettant" à Anthropic de larguer la grosse thunasse ne foncti… | ↗Not briefed |
bluesky bluesky-ai | 18d | Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts | Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts | ↗Not briefed |
bluesky bluesky-ai | 18d | Elon Musk pledged to award a share of SpaceX to every American child, part of what he has vowed as an effort to unleash “universal high inc… | Elon Musk pledged to award a share of SpaceX to every American child, part of what he has vowed as an effort to unleash “universal hig… | ↗Not briefed |
bluesky bluesky-ai | 18d | From what we can tell, yeah. The LLM is basically a glorified, environment-destroying search engine. Any script generation being done is ju… | From what we can tell, yeah. The LLM is basically a glorified, environment-destroying search engine. Any script generation being done… | ↗Not briefed |
arxiv arxiv-ai | 18d | Low Residuals, Wrong Value: Learning the Viscosity Solution of Hamilton-Jacobi-Isaacs Equations with PINNs | Physics-informed neural networks (PINNs) and other residual-trained solvers judge a candidate solution by the size of its sampled PDE… | ↗Not briefed |
arxiv arxiv-ai | 18d | CARLAR Influence: A Structured Hallucination Risk in Multimodal Cheque Extraction and an Exploratory Compressed-Image Countermeasure | Multimodal large language models (LLMs) are increasingly used for structured extraction from financial documents, but their behavior u… | ↗Not briefed |
rss techcrunch-ai | 18d | Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts | OpenCode is an open-source AI coding agent that's become a popular alternative to Claude Code and Cursor. But hackers got into part of… | ↗Not briefed |
hn hn-ai | 18d | Show HN: Maali – Privacy-first personal finance app with a local AI analyst | I never felt comfortable signing up for cloud budgeting apps and having them store all of my financial data on their servers. I also d… | ↗Not briefed |