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 | 1mo | New SDK for Photonic Compute? | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Diff Pro Max – File Compare App | An offline app for comparing differences in text, floders, images, PDF, Excel files.<p>Checkng for diffences is a task that comes up f… | ↗Not briefed |
bluesky bluesky-ai | 1mo | such a good benchmark lol | such a good benchmark lol | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM could have been massively successful if they kept expectations (and the actual output) in check with what it really is: arguably better… | LLM could have been massively successful if they kept expectations (and the actual output) in check with what it really is: arguably b… | ↗Not briefed |
hn hn-ai | 1mo | Slick: The most powerful and versatile Slack client mod | — | ↗Not briefed |
hn hn-ai | 1mo | The state of MCP server security, after reading thirteen of them | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A "talk to a real person now" video chat, self-hosted on Cloudflare | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Heads up to any fellow authors who are part of the claim against Anthropic. I just received an email from a dodgy lawyer firm trying to scr… | Heads up to any fellow authors who are part of the claim against Anthropic. I just received an email from a dodgy lawyer firm trying t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I’m sure they do! | I’m sure they do! | ↗Not briefed |
bluesky bluesky-ai | 1mo | I've heard some concern recently about "rogue AI" from friends a little less plugged into the discourse and this seems like a good read abo… | I've heard some concern recently about "rogue AI" from friends a little less plugged into the discourse and this seems like a good rea… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Apparently the CEO of Hoyo has stated that in the next couple years HoYoverse will “play one of the most pivotal roles in China’s LLM-AI” … | Apparently the CEO of Hoyo has stated that in the next couple years HoYoverse will “play one of the most pivotal roles in China’s LLM-… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 【2ch】deepseekなどの中華AIの中身は米国製「Claude」でアメリカに機密情報が漏れたことが判明 | 【2ch】deepseekなどの中華AIの中身は米国製「Claude」でアメリカに機密情報が漏れたことが判明 | ↗Not briefed |
bluesky bluesky-ai | 1mo | ‘Anthropic and OpenAI both released cheaper versions of their leading models on Tuesday, as the price war between the two leading US labs a… | ‘Anthropic and OpenAI both released cheaper versions of their leading models on Tuesday, as the price war between the two leading US l… | ↗Not briefed |
bluesky bluesky-ai | 1mo | De ene paragraaf ben je een leek, de andere weet je iedereen te vertellen dat OpenAI-frames de waarheid vertelt over hoe krachtig AI agents… | De ene paragraaf ben je een leek, de andere weet je iedereen te vertellen dat OpenAI-frames de waarheid vertelt over hoe krachtig AI a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Agent-Editing World Model: Rethinking World Modeling for LLM Agents | Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To furth… | ↗Not briefed |
hn hn-ai | 1mo | Pilgrim raises $25M seed for biological threat detection device | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Okay let's do this. I'm seeing people talking about this as proof that the LLM is doing genuine thinking and ideation. Or at the very least… | Okay let's do this. I'm seeing people talking about this as proof that the LLM is doing genuine thinking and ideation. Or at the very… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Frozen Flows Forget: Diagnosing and Restoring Lost Motion in a Latent-flow World Model | Latent world models that integrate a flow in a frozen self supervised latent space train stably and cheaply, yet silently lose the pro… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI researchers might be spending >$4M/day on tokens (at API prices) | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Diurnal – add a task without switching, with your calendar in view | Hi HN, I'm Anshul. Diurnal is a Mac to-do app I've been working on.<p>Many to-do apps have a quick-add shortcut: a text fiel… | ↗Not briefed |
rss simon-willison | 1mo | Gemini 3.8 TTS Playground | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | The whole debate about whether the AI execs even grasp what they're doing has been at least partially settled by the released documents in … | The whole debate about whether the AI execs even grasp what they're doing has been at least partially settled by the released document… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Artificial intelligence systems do not think, feel, want or understand. Avoid language that gives them human characteristics. | Artificial intelligence systems do not think, feel, want or understand. Avoid language that gives them human characteristics. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Holographic Reduced Representations with Clifford Variational Autoencoders | Vector Symbolic Algebras project data structures into a hyperdimensional vector space through the application of their vector algebras… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Collective Dynamics with Differentiable Gaussian Representations | Collective responses depend on individual differences, contact opportunities, and accumulated experience. Learning their dynamics from… | ↗Not briefed |
hn hn-ai | 1mo | Qwen Image 2.1 beats Google Nano Banana 2.0 with minuscule 7B parameter model | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Local outlets like ours operate on tight budgets and employ dedicated editorial staff who work every day to keep their communities informed… | Local outlets like ours operate on tight budgets and employ dedicated editorial staff who work every day to keep their communities inf… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Noisy Creek—publisher of the Chicago Reader, The Stranger in Seattle, and Portland Mercury—joined 60 publishers representing more than 550 … | Noisy Creek—publisher of the Chicago Reader, The Stranger in Seattle, and Portland Mercury—joined 60 publishers representing more than… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The Reader is joining a copyright lawsuit against OpenAI and Microsoft. 1/🧵 | The Reader is joining a copyright lawsuit against OpenAI and Microsoft. 1/🧵 | ↗Not briefed |
arxiv arxiv-ai | 1mo | Memory Attention | Language models typically construct attention values from contextual hidden states, even when some of their content may be reusable ac… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Fine-Tuning LLMs for Translation: General Forgetting Mitigation Does Not Preserve MT-Specific Instruction Following | Fine-tuning large language models on parallel data improves translation quality but can cause catastrophic forgetting. Mitigation meth… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Trying harder to articulate the difference: I think we consider technical skill and tool operation to not be art or worthy of protection, … | Trying harder to articulate the difference: I think we consider technical skill and tool operation to not be art or worthy of protect… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It is genuinely just silly for leading rationalists, effective altruists, and big AI companies to claim they're somehow totally separate en… | It is genuinely just silly for leading rationalists, effective altruists, and big AI companies to claim they're somehow totally separa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Quantum score matching with applications to learning thermal states | Score matching has driven major advances in classical generative learning by enabling models to learn from data without evaluating int… | ↗Not briefed |
arxiv arxiv-ai | 1mo | UltraBench 2: Towards Robust Evaluation of Vision Foundation Models on Ultrasound | Benchmarking is an increasingly critical part of research in machine learning and the domains where it is applied, including healthcar… | ↗Not briefed |
rss techcrunch-ai | 1mo | ChatGPT mobile app gets voice-based agentic features🔧 ChatGPT | Pro and Plus users will be able to use the Work tab on their phones | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence: Visually impaired hail AI smart glasses despite privacy concerns | A visually impaired man has hailed artificial intelligence-powered smart glasses as one of the greatest inventions of technology, desp… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence: Visually impaired hail AI smart glasses despite privacy concerns | A visually impaired man has hailed artificial intelligence-powered smart glasses as one of the greatest inventions of technology, desp… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence: Visually impaired hail AI smart glasses despite privacy concerns | A visually impaired man has hailed artificial intelligence-powered smart glasses as one of the greatest inventions of technology, desp… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence: Visually impaired hail AI smart glasses despite privacy concerns | A visually impaired man has hailed artificial intelligence-powered smart glasses as one of the greatest inventions of technology, desp… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Psychosis Guard – Safety for long LLM conversations | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I just think vibe-coding is like, the worst and laziest possible approach to LLM-assisted code. What happened to just using that kind of an… | I just think vibe-coding is like, the worst and laziest possible approach to LLM-assisted code. What happened to just using that kind… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In the same way that an LLM is racist because it is trained on racist material, the internet reproduces a great deal of racist ideas becaus… | In the same way that an LLM is racist because it is trained on racist material, the internet reproduces a great deal of racist ideas b… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When and Where to Trust the Teacher: Unifying On-Policy Distillation and GRPO through Entropy-Calibrated Credit Assignment | Reinforcement learning with verifiable rewards (RLVR) supervises mathematical reasoning through final-answer correctness, but provides… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ah, discourse on LLM cognition continues and I find myself mildly annoyed by a philosophy of mind paper as either trivial or potentially qu… | Ah, discourse on LLM cognition continues and I find myself mildly annoyed by a philosophy of mind paper as either trivial or potential… | ↗Not briefed |
bluesky bluesky-ai | 1mo | But what does it even mean to have an “original” insight? Anthropic already stole my book, which has the best of me in it. virginiaheff… | But what does it even mean to have an “original” insight? Anthropic already stole my book, which has the best of me in it. virgini… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Holden Karnofsky - co-founder of GiveWell, co-founder and former CEO of Open Philanthropy/Coefficient Giving, husband of Anthropic Presiden… | Holden Karnofsky - co-founder of GiveWell, co-founder and former CEO of Open Philanthropy/Coefficient Giving, husband of Anthropic Pre… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Sanemark – Markdown LSP for notes readable in plain text | Hey HN!<p>I built Sanemark, a Markdown LSP for taking notes.<p>Some design choices:<p>1. Automatically moves inline links to reference… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ForgetMimic: Motion Unlearning for Reinforcement Learning Humanoid Control | Humanoid control, leveraging human demonstrations, has achieved diverse, agile, and natural locomotion behaviors through reinforcement… | ↗Not briefed |
hn hn-ai | 1mo | Serve DiffusionGemma-Jev (Djev) on a TypeSafe AI Compatible API | — | ↗Not briefed |