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 | 1mo | As AI tools become more prolific, I want folks to remember these are not HIPAA-protected platforms. Your sensitive health data is YOURS al…🔧 ChatGPT | As AI tools become more prolific, I want folks to remember these are not HIPAA-protected platforms. Your sensitive health data is YOU… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wow incredible, OpenAI has somehow come up with an even worse solution to the problem of “we will brute force all pure mathematics and ruin… | Wow incredible, OpenAI has somehow come up with an even worse solution to the problem of “we will brute force all pure mathematics and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Reading their docs, what they're building looks a whole like what ML looked like pre-LLM explosion? It'd be cool if this means "building … | Reading their docs, what they're building looks a whole like what ML looked like pre-LLM explosion? It'd be cool if this means "buil… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Reading the press about Jev and TypeSafe AI and it's wild to see ppl seemingly re-discovering what folks like @timnitgebru.blacksky.app hav… | Reading the press about Jev and TypeSafe AI and it's wild to see ppl seemingly re-discovering what folks like @timnitgebru.blacksky.ap… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI can be a really wonderful tool for exploring topics far from coding. I had Claude Fable 5.1 put together an annotated guide to Eliot's … | AI can be a really wonderful tool for exploring topics far from coding. I had Claude Fable 5.1 put together an annotated guide to Eli… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 中国AIのKimi、米Claudeに｢代理回答｣させる アンソロピックが指摘 - 日本経済新聞 www.nikkei.com/article/DGXZ... | 中国AIのKimi、米Claudeに｢代理回答｣させる アンソロピックが指摘 - 日本経済新聞 www.nikkei.com/article/DGXZ... | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is the kind of question that would be good to ask a TI-84 and an LLM | This is the kind of question that would be good to ask a TI-84 and an LLM | ↗Not briefed |
bluesky bluesky-ai | 1mo | https://www.theinformation.com/briefings/alibaba-appoints-new-head-qwen-llm-team | https://www.theinformation.com/briefings/alibaba-appoints-new-head-qwen-llm-team | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Defensive Policies against Diverse Inference Attacks for Smart Meter Privacy | Smart meter (SM) data provides fine-grained visibility into household energy consumption, but also exposes users to privacy risks. Inf… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I assume we'll get a "no true RSI" sort of like AGI. I think "claude leads 25% of R&D projects in anthropic" is already RSI by many reasona… | I assume we'll get a "no true RSI" sort of like AGI. I think "claude leads 25% of R&D projects in anthropic" is already RSI by many re… | ↗Not briefed |
bluesky bluesky-ai | 1mo | というか、ぶっちゃけLLM基盤が「著作物をさんざん無断学習した結果」で、これ自体「国際標準でOKなの？」って突っ込みたい。 まずそこが歪んでるから諸問題が起こるのでは？ | というか、ぶっちゃけLLM基盤が「著作物をさんざん無断学習した結果」で、これ自体「国際標準でOKなの？」って突っ込みたい。 まずそこが歪んでるから諸問題が起こるのでは？ | ↗Not briefed |
bluesky bluesky-ai | 1mo | Slop in my desktop, slop in my phone, slop in my games, slop in my TV, slop in my job, slop in healthcare... SLOP SLOP SLOP www.theregiste… | Slop in my desktop, slop in my phone, slop in my games, slop in my TV, slop in my job, slop in healthcare... SLOP SLOP SLOP www.there… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Terminal Shrinkage Averaging Reveals a Schedule-Estimator Interaction in LLM Pretraining | Large language model (LLM) pretraining conventionally returns the raw final iterate. This couples two design choices: the learning-rat… | ↗Not briefed |
bluesky bluesky-ai | 1mo | They didn’t invent AI to make us stupid and illiterate they invented it so that the military can drop ebola bombs on all of mongolia and sa… | They didn’t invent AI to make us stupid and illiterate they invented it so that the military can drop ebola bombs on all of mongolia a… | ↗Not briefed |
hn hn-ai | 1mo | Looking forward to Git 2.56 – and 3.0 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Put together some notes on Jev and the new category of system one aka decision models simonwillison.net/2026/Sep/21/... | Put together some notes on Jev and the new category of system one aka decision models simonwillison.net/2026/Sep/21/... | ↗Not briefed |
bluesky bluesky-ai | 1mo | maybe we could talk about how drag'n wash devs used claude to write at least some of the game's code while incorrectly claiming that 'no ge… | maybe we could talk about how drag'n wash devs used claude to write at least some of the game's code while incorrectly claiming that '… | ↗Not briefed |
bluesky bluesky-ai | 1mo | You could do a LLM very slowly *by hand*. The math is the only necessary component. Everything else you listed is just for efficiency gai… | You could do a LLM very slowly *by hand*. The math is the only necessary component. Everything else you listed is just for efficienc… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Building Seven in Public | <a href="https://github.com/02zerocool/truth-benchmark" rel="nofollow">https://github.com/02zerocoo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I tested Xiaomi MiMo-V2.6 Pro on my LLM benchmark and it is the highest scoring open-weight model seen so far. Beats Muse Spark 1.3 on pric… | I tested Xiaomi MiMo-V2.6 Pro on my LLM benchmark and it is the highest scoring open-weight model seen so far. Beats Muse Spark 1.3 on… | ↗Not briefed |
rss simon-willison | 1mo | Jev introduces a new shape of LLM - System One, aka Decision Models | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | After OpenAI Cyberattack, UN Panel Warns Current Guardrails Are 'Unraveling' www.commondreams.org/news/united-... | After OpenAI Cyberattack, UN Panel Warns Current Guardrails Are 'Unraveling' www.commondreams.org/news/united-... | ↗Not briefed |
bluesky bluesky-ai | 1mo | #Jev, a new #AI decision making model from #TypeSafeAI, answers questions in under half a second for a fraction of the cost of traditional …🔧 LangChain | #Jev, a new #AI decision making model from #TypeSafeAI, answers questions in under half a second for a fraction of the cost of traditi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ⚖️ CBC reports B.C. is suing OpenAI, alleging ChatGPT threats linked to the Tumbler Ridge shooter triggered an internal review but weren’t …🔧 ChatGPT | ⚖️ CBC reports B.C. is suing OpenAI, alleging ChatGPT threats linked to the Tumbler Ridge shooter triggered an internal review but wer… | ↗Not briefed |
rss techmeme | 1mo | A researcher says a flaw in Meta's Muse app for Mac lets any app or terminal command gain access to the token that authenticates users to their Muse account (Dan Goodin/Ars Technica) | Dan Goodin / Ars Technica: A researcher says a flaw in Meta's Muse app for Mac lets any app or terminal command gain access to the tok… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Spectra: A Rules-Driven LLM Pipeline for Automated KYC Document Processing | Know Your Client (KYC) onboarding in capital markets requires analysts to manually classify documents, extract structured data from he… | ↗Not briefed |
bluesky bluesky-ai | 1mo | British Columbia sues OpenAI for alleged safety violations and negligence for failing to flag the Tumbler Ridge shooting suspect's ChatGPT …🔧 ChatGPT | British Columbia sues OpenAI for alleged safety violations and negligence for failing to flag the Tumbler Ridge shooting suspect's Cha… | ↗Not briefed |
rss techmeme | 1mo | British Columbia sues OpenAI for alleged safety violations and negligence for failing to flag the Tumbler Ridge shooting suspect's ChatGPT activity to police (Georgia Wells/Wall Street Journal)🔧 ChatGPT | Georgia Wells / Wall Street Journal: British Columbia sues OpenAI for alleged safety violations and negligence for failing to flag the… | ↗Not briefed |
hn hn-ai | 1mo | LLMs Are Too Big. My Log Router Doesn't Need to Sing | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Practical Recipe for Semi-Supervised Federated ASR: Online Pseudo-Labels with Server Update Stabilization | Semi-supervised federated learning (SSFL) trains models on clients' unlabeled data using a teacher to generate pseudo-labels, with a s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Double digit commits are pushing it, but are still outweighed by human input 9 times out of 10 Triple Digit commits and up are where the L… | Double digit commits are pushing it, but are still outweighed by human input 9 times out of 10 Triple Digit commits and up are where… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RAG-NAROK: Retrieval-Aware Knowledge Corpus Poisoning in RAG with Source-specific Refutation | Retrieval augmented generation (RAG) systems have emerged as the dominant architecture for grounding large language model (LLM) output… | ↗Not briefed |
bluesky bluesky-ai | 1mo | As if we needed another reason to be concerned about OpenAI and Anthropic, they're now MAHA corporate sponsors. At this rate, AI should ha… | As if we needed another reason to be concerned about OpenAI and Anthropic, they're now MAHA corporate sponsors. At this rate, AI shou… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ShowTellArena: Evaluating Business Workflow Understanding from Demonstrations | We often teach a colleague by showing the work and explaining the decisions as we go. How can we check what an agent understood from t… | ↗Not briefed |
hn hn-ai | 1mo | Self-hosted, end-to-end encrypted 1:1 chat app. Offline key-exchange | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Real-Time Hand Gesture Recognition for OpenXR Using Transformer-Based Machine Learning | Hand gesture recognition is a key component in human-computer interaction (HCI), enabling intuitive interfaces for applications in gam… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Transom – easily manage multiple browser profiles on macOS | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | One of the funniest things about pro-LLM people is they always have to move to hypotheticals rather than the reality in front of them. LLM… | One of the funniest things about pro-LLM people is they always have to move to hypotheticals rather than the reality in front of them.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Like trivially you can think "LLMs do *not* have mental states currently" and "it is possible for future LLM-like entities to at some point… | Like trivially you can think "LLMs do *not* have mental states currently" and "it is possible for future LLM-like entities to at some… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rollout Efficiency in Reinforcement Learning for Reasoning Large Language Models: A Taxonomy and Future Directions | Reasoning-oriented reinforcement learning enables large language models to solve mathematical, coding, and other multi-step tasks, but… | ↗Not briefed |
hn hn-ai | 1mo | An LLM Beat NetHack | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic-linked CVEs pile up, attackers mostly shrug | Anthropic-linked CVEs pile up, attackers mostly shrug | ↗Not briefed |
bluesky bluesky-ai | 1mo | fr they are so cool for so many things but honestly if you're a pure vibecoder, for reasons like this, the LLM is a better tool for *learni… | fr they are so cool for so many things but honestly if you're a pure vibecoder, for reasons like this, the LLM is a better tool for *l… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If an LLM says, "I'm going to write code that does X," and sends code to a compiler which successfully does X, you are contending that the … | If an LLM says, "I'm going to write code that does X," and sends code to a compiler which successfully does X, you are contending that… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Transformer-Informed Trajectory Optimization for Relative Motion in Cislunar Orbits | Autonomous spacecraft guidance and control requires a fast solution to non-convex trajectory optimization, which can be accelerated by… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I hate that I can't trust ANY new 3D / raycaster NES project that pops up. Whenever I look deeper, it's always tainted by Claude AI. Every… | I hate that I can't trust ANY new 3D / raycaster NES project that pops up. Whenever I look deeper, it's always tainted by Claude AI.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this is the thing, long before AI the philosophers looking at this stuff have repeatedly chosen to basically minimally engage at best with … | this is the thing, long before AI the philosophers looking at this stuff have repeatedly chosen to basically minimally engage at best… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is why interpretability is so vital. We don't really know how models function beyond the transformer block. Once you get into the feed… | This is why interpretability is so vital. We don't really know how models function beyond the transformer block. Once you get into the… | ↗Not briefed |
hn hn-ai | 1mo | Almost started a war: Report reveals dark reality as AI permeates civilisation | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sara Nelson's new websites are a grim example of what happens when an organization replaces human artistic/design judgment with AI. Let an… | Sara Nelson's new websites are a grim example of what happens when an organization replaces human artistic/design judgment with AI. L… | ↗Not briefed |