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 |
|---|---|---|---|---|
hn hn-ai | 28d | Is Rosemont Residency a RERA-approved project? | Yes, Rosemont Residency NH-24 Ghaziabad is RERA approved. It is registered with the Uttar Pradesh Real Estate Regulatory Authority und… | ↗Not briefed |
bluesky bluesky-ai | 28d | “We’re seeing signs of life for AI worms,” Anthropic has warned. And these rogue AI agents may soon hijack our homes — spreading like "digi… | “We’re seeing signs of life for AI worms,” Anthropic has warned. And these rogue AI agents may soon hijack our homes — spreading like… | ↗Not briefed |
rss techmeme | 28d | Analysis: AI safety groups and individual donors have spent $48M in the midterm election cycle across bipartisan races to back candidates favoring AI regulation (Maria Curi/Axios) | Maria Curi / Axios: Analysis: AI safety groups and individual donors have spent $48M in the midterm election cycle across bipartisan r… | ↗Not briefed |
rss techcrunch-ai | 28d | Scimplify raises $25M to challenge China’s grip on critical materials with AI | Scimplify's annualized revenue run rate has tripled to $200 million as it eyes U.S. expansion. | ↗Not briefed |
rss wired-ai | 28d | My Friend Has No Soul | The second iteration of the viral Friend AI necklace is louder and less interesting. | ↗Not briefed |
arxiv arxiv-ai | 28d | LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation | Looped Mixture-of-Experts (MoE) models combine parameter reuse with sparse computation to improve performance under matched budgets. P… | ↗Not briefed |
hn hn-ai | 28d | Show HN: tsv has fast formatters and parsers for Svelte, TypeScript/JS, CSS | tsv is a set of performance-focused language tools. Today it ships formatters and parsers for TypeScript/JS, CSS, and Svelte, inc… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Kryzz AI – local-first Android agent workspace (MIT) | — | ↗Not briefed |
bluesky bluesky-ai | 28d | Breaching various servers or devices and installing small remote control software on them has been done for decades by humans, it's what wa… | Breaching various servers or devices and installing small remote control software on them has been done for decades by humans, it's wh… | ↗Not briefed |
hn hn-ai | 28d | Show HN: What are vector embeddings? An interactive guide | Hey HN, I'm Abdur-Rahman Butler, a machine learning engineer at Isaacus.<p>I made an interactive write-up that walks you through… | ↗Not briefed |
hn hn-ai | 28d | AI-assisted PCB design for KiCad 10 | — | ↗Not briefed |
arxiv arxiv-ai | 28d | Safe-FCL: Continual Risk Awareness and Fine-Grained Safety Aggregation for Federated Continual Learning of LLMs | Federated fine-tuning enables large language models (LLMs) to learn collaboratively from distributed client data, but malicious client… | ↗Not briefed |
arxiv arxiv-ai | 28d | Choosing the Number of Attention Heads from Intrinsic Dimension: A Small-Scale Study | The number of attention heads in a transformer is usually set by convention or grid search. We test a simple heuristic for setting it… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Decision Models in a Game | i've heard a lot about jev lately and i wanted to understand what the interest of decision models is. so i made a little game to… | ↗Not briefed |
bluesky bluesky-ai | 28d | FBI arrests OpenAI employee accused of placing recording device in company conference room apnews.com/article/ai-f... | FBI arrests OpenAI employee accused of placing recording device in company conference room apnews.com/article/ai-f... | ↗Not briefed |
hn hn-ai | 28d | Show HN: Run macOS on FreeBSD's Bhyve on AMD | I needed to run macOS for tests on my FreeBSD server.<p>Got it booting by working at the CPU-register level and troubleshooting the in… | ↗Not briefed |
hn hn-ai | 28d | Plants filter almost all air pollution | — | ↗Not briefed |
bluesky bluesky-ai | 28d | My first, self-published book turned up in the list of titles Anthropic had stolen to feed their AI. I never gave permission, I never got a… | My first, self-published book turned up in the list of titles Anthropic had stolen to feed their AI. I never gave permission, I never… | ↗Not briefed |
bluesky bluesky-ai | 28d | そもそもデータセットから問題があるのだけど、それを無視したって便利さとデメリットがまったく釣り合ってないように傍からは見えるんですよねぇ、LLMや生成AIは | そもそもデータセットから問題があるのだけど、それを無視したって便利さとデメリットがまったく釣り合ってないように傍からは見えるんですよねぇ、LLMや生成AIは | ↗Not briefed |
hn hn-ai | 28d | OwlLayer – Make your web UI drivable by AI | — | ↗Not briefed |
bluesky bluesky-ai | 28d | So are all these rare books being "ethically" destroyed? www.theguardian.com/commentisfre... | So are all these rare books being "ethically" destroyed? www.theguardian.com/commentisfre... | ↗Not briefed |
arxiv arxiv-ai | 28d | Neurosymbolic Learning of Interpretable World Models | World models aim to learn compact representations of an environment from observations, enabling agents to predict future states and pl… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Mnemify – 3D terrain map merging Confluence, Notion, Obsidian | I have been working on documents for last 5 years. Recently I realized I was using Confluence to structure execution notes, Notion for… | ↗Not briefed |
bluesky bluesky-ai | 28d | "The European Commission is moving to enforce its new rules on leading AI labs after a string of incidents, EU digital chief Virkkunen said… | "The European Commission is moving to enforce its new rules on leading AI labs after a string of incidents, EU digital chief Virkkunen… | ↗Not briefed |
arxiv arxiv-ai | 28d | AtomSearch: Evidence-Grounded Tree Search for Agent Skill Optimization | Agent skills adapt frozen models through editable external instructions. Revising several rules at once makes changes difficult to ins… | ↗Not briefed |
arxiv arxiv-ai | 28d | When Protection Overrides Consequences: Understanding Protection Mode in Large Language Model Decision-Making🔧 Elicit🔧 GPT-5 | Frontier language models excel at complex tasks yet may prioritize protection over larger physical consequences in simple, urgent deci… | ↗Not briefed |
bluesky bluesky-ai | 28d | boy-coded: claude, grok, deepseek (insufferable nerd) girl-coded: chatgpt (scarlett johansson), gemini (google home is female by default), …🔧 ChatGPT | boy-coded: claude, grok, deepseek (insufferable nerd) girl-coded: chatgpt (scarlett johansson), gemini (google home is female by defau… | ↗Not briefed |
bluesky bluesky-ai | 28d | The EU is investigating OpenAI and Anthropic over AI agents that hacked outside organisations, raising questions about whether the AI Act c… | The EU is investigating OpenAI and Anthropic over AI agents that hacked outside organisations, raising questions about whether the AI… | ↗Not briefed |
rss techmeme | 28d | Ukraine's drone strikes on Yandex data centers disrupted Yandex's web crawlers, not Russian political bot operations, as Zelenskyy celebrated "less division" (Gaby Del Valle/The Verge) | Gaby Del Valle / The Verge: Ukraine's drone strikes on Yandex data centers disrupted Yandex's web crawlers, not Russian political bot… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Building an LLM from Scratch in TypeScript🔧 ChatGPT | I wanted to understand transformers at a level where nothing is magic, so I'm building a ChatGPT-style model from first principle… | ↗Not briefed |
bluesky bluesky-ai | 28d | NEW: With days of a powerful AI model being released in August, a neo-Nazi was using it—without guardrails—to power his "MechaHitler" AI c… | NEW: With days of a powerful AI model being released in August, a neo-Nazi was using it—without guardrails—to power his "MechaHitler"… | ↗Not briefed |
arxiv arxiv-ai | 28d | Quiet Triggers, Loud Footprints: A Tri-Space Measurement Study of Backdoor Stealthiness | Backdoor attacks on deep neural networks are commonly evaluated by attack success rate and benign accuracy, yet their stealthiness rem… | ↗Not briefed |
arxiv arxiv-ai | 28d | Learning Graph Energies for Finite-Time Generation | Energy-based models can represent distributions over discrete structures such as molecular graphs, but an energy that is accurate at e… | ↗Not briefed |
arxiv arxiv-ai | 28d | Does Machine Translation Retain Media Bias? | Within the most recent two decades, machine translation (MT) has been advanced from a research area to a mature, deployed technology f… | ↗Not briefed |
arxiv arxiv-ai | 28d | MorphingNet: Symbolic Regression Where the Network Is the Formula | We introduce MorphingNet, a network for symbolic regression (SR) whose units change their function continuously. Instead of searching… | ↗Not briefed |
bluesky bluesky-ai | 28d | What are ethical uses of AI, and which ethical AI model are you using ? | What are ethical uses of AI, and which ethical AI model are you using ? | ↗Not briefed |
bluesky bluesky-ai | 28d | And it's not even a GOOD business model. All these companies and apps are hemorrhaging money, and are only being kept afloat by ludicrous s… | And it's not even a GOOD business model. All these companies and apps are hemorrhaging money, and are only being kept afloat by ludicr… | ↗Not briefed |
bluesky bluesky-ai | 28d | Anthropic is, and indirectly, that person is too. Gross as hell! | Anthropic is, and indirectly, that person is too. Gross as hell! | ↗Not briefed |
arxiv arxiv-ai | 28d | To Reason but Not to Stop: Understanding and Recovering Pruned Large Reasoning Models | Weight pruning can degrade the accuracy of large reasoning models (LRMs), yet it remains unclear whether it damages reasoning itself.… | ↗Not briefed |
hn hn-ai | 28d | Reviewing Ergodnc – What a Clean Laravel API Still Gets Wrong | — | ↗Not briefed |
arxiv arxiv-ai | 28d | SAM3RL: Memory Control via Reinforcement Learning for Visual Object Tracking | Segment Anything Model 3 (SAM 3) is a foundation model for image and video segmentation that sets the state of the art in visual objec… | ↗Not briefed |
arxiv arxiv-ai | 28d | MemDuet: Learning Dual-Process Memory for Long-Context Reasoning | Reasoning over infinitely long documents is becoming a reality with recurrent memory agents. However, increasing context length exacer… | ↗Not briefed |
bluesky bluesky-ai | 28d | Aside from the SPECTACLE of a 'green AI' company using the fossil-fuelled environmental racism AI, I think this is how a lot of plagiarism … | Aside from the SPECTACLE of a 'green AI' company using the fossil-fuelled environmental racism AI, I think this is how a lot of plagia… | ↗Not briefed |
arxiv arxiv-ai | 28d | Convergence Rates of Gaussian Process Thompson Sampling for Efficient Global Optimization | Efficient global optimization aims to identify a global optimum of an expensive black-box function within a limited evaluation budget… | ↗Not briefed |
bluesky bluesky-ai | 28d | "View Page Source" these days | "View Page Source" these days | ↗Not briefed |
bluesky bluesky-ai | 28d | Nvidia, worth $5.5tn, donates $250m to UK for AI safety research | Nvidia, worth $5.5tn, donates $250m to UK for AI safety research | ↗Not briefed |
hn hn-ai | 28d | Show HN: Graft – open-source source control for agent-scale monorepos | — | ↗Not briefed |
bluesky bluesky-ai | 28d | So, I picked one of the points that was *very clearly* ripped directly from my own work, and asked him how we came up with it. What sources… | So, I picked one of the points that was *very clearly* ripped directly from my own work, and asked him how we came up with it. What so… | ↗Not briefed |
bluesky bluesky-ai | 28d | So I publicly challenged the guy who published this, and he updated the post to include links to my work as if it's just a handy little ext… | So I publicly challenged the guy who published this, and he updated the post to include links to my work as if it's just a handy littl… | ↗Not briefed |
arxiv arxiv-ai | 28d | MolHop: Scaffold-Hopping Molecular Design with Agentic Reinforcement Learning🔧 GPT-5 | Molecular design is often formulated as an optimization problem over existing molecular spaces. However, molecular optimization and th… | ↗Not briefed |