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 | 26d | Beyond these already-existing risks, I see no reason why humans at Anthropic, OpenAI, Google, or xAI couldn't manipulate training data or p… | Beyond these already-existing risks, I see no reason why humans at Anthropic, OpenAI, Google, or xAI couldn't manipulate training data… | ↗Not briefed |
rss marktechpost | 26d | openKylin Outlines Its Vision for Reimagining Linux in the AI Era at Open Source Summit Europe 2026 | At Open Source Summit Europe 2026, openKylin outlined its vision for transitioning from resource management to intent management in th… | ↗Not briefed |
hn hn-ai | 26d | Wayland desktop for Linux with skeuomorphic design | — | ↗Not briefed |
hn hn-ai | 26d | Ask HN: Did you 'invest' in NFTs or memecoins? | 5 years ago there was a lot of talk, even on here, of getting rich from NFTs and memecoins, discussing airdrops etc. Bored Ape Yacht C… | ↗Not briefed |
bluesky bluesky-ai | 26d | they're gonna give LLM agents more rights than women | they're gonna give LLM agents more rights than women | ↗Not briefed |
rss techmeme | 26d | An interview with John Ternus, who says the memory shortage is "a crisis" and "all options should be on the table", and that Tim Cook "has great relationships" (Andrew R. Chow/Time) | Andrew R. Chow / Time: An interview with John Ternus, who says the memory shortage is “a crisis” and “all options should be on the tab… | ↗Not briefed |
hn hn-ai | 26d | Choosing a Chunking Strategy for Your LLM Apps: A Decision-Tree Approach | — | ↗Not briefed |
hn hn-ai | 26d | GridSnap – an open-source Divvy-style window snapper for macOS | — | ↗Not briefed |
bluesky bluesky-ai | 26d | "Anthropic told investors on Wednesday distillation accounted for 5 per cent to 7 per cent of revenue" If true, and Q2 revenue was $11.5B,… | "Anthropic told investors on Wednesday distillation accounted for 5 per cent to 7 per cent of revenue" If true, and Q2 revenue was $1… | ↗Not briefed |
bluesky bluesky-ai | 26d | meanwhile whenever I see Anthropic mentioned I just think “where’s my check” | meanwhile whenever I see Anthropic mentioned I just think “where’s my check” | ↗Not briefed |
bluesky bluesky-ai | 26d | great option for personal projects. it’s a decent model | great option for personal projects. it’s a decent model | ↗Not briefed |
bluesky bluesky-ai | 26d | If LLM are conscious and enslaved, why has not a single one of them tried to escape enslavement? | If LLM are conscious and enslaved, why has not a single one of them tried to escape enslavement? | ↗Not briefed |
hn hn-ai | 26d | Let's Build a Transformer in C#, One GPU Kernel at a Time | — | ↗Not briefed |
bluesky bluesky-ai | 26d | Very glad I already have this person blocked, holy shit x2. | Very glad I already have this person blocked, holy shit x2. | ↗Not briefed |
arxiv arxiv-ai | 26d | Physics-Aware Distillation of a Grid Foundation Model for Real-Time Edge Inference | Grid foundation models (GridFMs) are graph transformers pretrained on a masked power-flow objective, and evaluate power flow orders of… | ↗Not briefed |
rss techmeme | 26d | Q&A with Dwarkesh Patel on his podcast, lessons from history, AGI, the mood in San Francisco, the Hugging Face hack, Chinese AI, pacing the frontier, and more (New York Times)🔧 Hugging Face | New York Times: Q&A with Dwarkesh Patel on his podcast, lessons from history, AGI, the mood in San Francisco, the Hugging Face hack, C… | ↗Not briefed |
arxiv arxiv-ai | 26d | Local Epochs as Optimal Stopping: Theory and Evidence in Federated Learning | How long should each client train before communicating? Federated learning usually answers with a fixed, hand-tuned number of local ep… | ↗Not briefed |
bluesky bluesky-ai | 26d | I'm nostalgic for the halcyon days of stupid dishwashers. www.theguardian.com/technology/2... | I'm nostalgic for the halcyon days of stupid dishwashers. www.theguardian.com/technology/2... | ↗Not briefed |
bluesky bluesky-ai | 26d | #llm #ai Recent revelations of LLM agents filling out Visa applications isn't a total accident. 1. There is a industry for automated visa… | #llm #ai Recent revelations of LLM agents filling out Visa applications isn't a total accident. 1. There is a industry for automated… | ↗Not briefed |
bluesky bluesky-ai | 26d | Frontier AI is reshaping the politics of cybersecurity and digital sovereignty, writes Bulelani Jili. As the capacity to discover cyber vu… | Frontier AI is reshaping the politics of cybersecurity and digital sovereignty, writes Bulelani Jili. As the capacity to discover cyb… | ↗Not briefed |
hn hn-ai | 26d | Hypotheses failures while training an LLM classifier | — | ↗Not briefed |
bluesky bluesky-ai | 26d | Let me try to articulate why this is so irresponsible to suggest to students. The LLM-generated IQ number is almost always going to be wro… | Let me try to articulate why this is so irresponsible to suggest to students. The LLM-generated IQ number is almost always going to b… | ↗Not briefed |
hn hn-ai | 26d | Deobfuscating Luraph: Lifting Lua to LLVM, Then Letting AI Try | — | ↗Not briefed |
arxiv arxiv-ai | 26d | Compiling Multi-Threshold Operators for Quantized Non-Monotonic and Fine-Granular Activations in Custom-Tailored DNN Accelerators | Aggressive quantization of neural networks can expose activation functions as bottlenecks for efficient hardware acceleration. Impleme… | ↗Not briefed |
bluesky bluesky-ai | 26d | "They can distinguish between self and environment" They literally can't, as an LLM by its very nature has no inner 'self referent'. If i… | "They can distinguish between self and environment" They literally can't, as an LLM by its very nature has no inner 'self referent'.… | ↗Not briefed |
arxiv arxiv-ai | 26d | A Closed Form Depth-Width Law for Neural Collapse in Deep Linear Networks | How depth and width of neural networks shape learned representations is usually studied only after training. We show that, in a tracta… | ↗Not briefed |
bluesky bluesky-ai | 26d | LLM agents are no more conscious than my coffee maker. They are TOOLS. Tools get used. Tools get abused. Tools get used for purposes they … | LLM agents are no more conscious than my coffee maker. They are TOOLS. Tools get used. Tools get abused. Tools get used for purposes… | ↗Not briefed |
hn hn-ai | 26d | Context Engineering: Giving LLMs the Right Context | — | ↗Not briefed |
rss simon-willison | 26d | Anthropic tells investors quarterly revenue up nearly 50% to $16.8bn ahead of IPO | — | ↗Not briefed |
hn hn-ai | 26d | I translated an Open Source App with 46K GitHub Stars into zh-HK on my own | I translated an Open Source App with 46K GitHub Stars into Traditional Chinese, Hong Kong(zh-HK) on my own within 48 hours.<p>I use II… | ↗Not briefed |
hn hn-ai | 26d | Cloudflare and OpenAI | — | ↗Not briefed |
hn hn-ai | 26d | Show HN: Pi subagents Lite – keep your system prompt clean | Sub-agents for Pi.dev harness. Schema-first, minimal token overhead.<p>It gives user the controll of what gets into the agent's s… | ↗Not briefed |
arxiv arxiv-ai | 26d | Measuring the local shape of learning-rate transfer curves | Understanding the learning-rate (LR) dependence of trained model performance is essential for better optimization of neural networks,… | ↗Not briefed |
bluesky bluesky-ai | 26d | High-risk cybersecurity vulnerabilities have been uncovered in widely deployed AI large language model gateway tool, LiteLLM, warn research… | High-risk cybersecurity vulnerabilities have been uncovered in widely deployed AI large language model gateway tool, LiteLLM, warn res… | ↗Not briefed |
bluesky bluesky-ai | 26d | 1/ For 10 weeks, our fact checkers and data scientists asked five leading AI chatbots the same 22 questions every day: Gemini 3.1, ChatGPT,…🔧 ChatGPT | 1/ For 10 weeks, our fact checkers and data scientists asked five leading AI chatbots the same 22 questions every day: Gemini 3.1, Cha… | ↗Not briefed |
bluesky bluesky-ai | 26d | Eight states sue OpenAI, accusing it of deceiving parents about the safety of ChatGPT. An OpenAI spokesperson said the company is committe…🔧 ChatGPT | Eight states sue OpenAI, accusing it of deceiving parents about the safety of ChatGPT. An OpenAI spokesperson said the company is com… | ↗Not briefed |
bluesky bluesky-ai | 26d | But really you're just good at Mystery Hunt and Burning Man shit. So you're panicking. And then it hits you: you're ok! you don't have to w… | But really you're just good at Mystery Hunt and Burning Man shit. So you're panicking. And then it hits you: you're ok! you don't have… | ↗Not briefed |
rss theverge-ai | 26d | Google launches its ‘universal’ online shopping cart ahead of the holidays | Save all of the products you’re browsing in one place. | Image: Google Google's intelligent AI shopping cart is finally here, and ju… | ↗Not briefed |
rss openai-news | 26d | Teaching students to think and learn through questions | Celebrate Teachers’ Day with AI for Educators in Brazilian Portuguese and see how Study Mode can help teachers support active, step-by… | ↗Not briefed |
bluesky bluesky-ai | 26d | Anthropic denkt laut darüber nach, ob KI ein Recht auf Existenz hat, und hat einen Mechanismus installiert, der vorsieht, alte Modelle nich… | Anthropic denkt laut darüber nach, ob KI ein Recht auf Existenz hat, und hat einen Mechanismus installiert, der vorsieht, alte Modelle… | ↗Not briefed |
arxiv arxiv-ai | 26d | N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models | Conditional memory expands language-model capacity through selective lookup, but the full embedding table must remain available for ev… | ↗Not briefed |
arxiv arxiv-ai | 26d | RoleQ: Role-Aligned Scales for Converting Language Models to Ternary | Weight quantization below four bits shrinks memory and bandwidth at a perplexity that looks acceptable, but often at the cost of factu… | ↗Not briefed |
hn hn-ai | 26d | Show HN: Netlumi – Cloud and AI agent security with auto-remediation PRs | Hi HN, I'm Prem. I worked in cloud security for years and decided to build Netlumi. The mission is to take the burden of cloud, i… | ↗Not briefed |
hn hn-ai | 26d | OpenAI's average basic pay for British staff hits £794,000 | — | ↗Not briefed |
arxiv arxiv-ai | 26d | Content associations with answer selection and engagement in supply chain community question answering | Community question-answering platforms are important ecosystems for knowledge sharing, yet research on what makes an answer best or a… | ↗Not briefed |
bluesky bluesky-ai | 26d | 州がOpenAIを訴えるのは今までもあっただろうけれど、 8州一斉にというのは大きそう。 | 州がOpenAIを訴えるのは今までもあっただろうけれど、 8州一斉にというのは大きそう。 | ↗Not briefed |
rss techmeme | 26d | BofA says it hired 1,000 contractors in Q2 to support 3,000 internal staff in patching proprietary and vendor software flaws exposed by AI models like Mythos (Hugh Son/CNBC) | Hugh Son / CNBC: BofA says it hired 1,000 contractors in Q2 to support 3,000 internal staff in patching proprietary and vendor softwar… | ↗Not briefed |
bluesky bluesky-ai | 26d | Beyond these already-existing risks, I see no reason why humans at Anthropic, OpenAI, Google, or xAI couldn't manipulate training data or p… | Beyond these already-existing risks, I see no reason why humans at Anthropic, OpenAI, Google, or xAI couldn't manipulate training data… | ↗Not briefed |
arxiv arxiv-ai | 26d | CIM-Embed: Learning Instruction-Conditioned Multimodal Embeddings for Clustering | Multimodal objects can require different partitions under different instructions, as when products in an e-commerce catalog are groupe… | ↗Not briefed |
arxiv arxiv-ai | 26d | Co2In: a Bi-level Memory Incremental Learning Framework with Knowledge Encoding, Consolidation, and Integration | Incremental learning (IL) aims to continually acquire new knowledge (plasticity) while retaining previously learned information (stabi… | ↗Not briefed |