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 | 2mo | Flushing toilet releases 'plume' of breathable pathogens – closing lid helps | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | Disentangling Representation Evolution in Transformers through Directional Decomposition | Transformer representations evolve through learned additive transformations that either preserve their current direction or redirect i… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Discovery Foundation Models: Toward Open-Ended Discovery Intelligence | Foundation models have progressed from learning and reasoning over existing knowledge, to increasingly learning through action, tool u… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Mind2Dialogue: Training Human-Aware Language Models by Simulating User Mental States | As language models become more capable, long-term collaboration in learning, reasoning, and decision-making calls for a deeper underst… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Stasi 2.0: OpenAI lässt andere Menschen DEINEN Chat mitlesen. An lokaler KI führt kein Weg mehr vorbei. Kein Wunder dass die KI Bros ih… | Stasi 2.0: OpenAI lässt andere Menschen DEINEN Chat mitlesen. An lokaler KI führt kein Weg mehr vorbei. Kein Wunder dass die KI Br… | ↗Not briefed |
rss mit-tech-review-ai | 2mo | The AI industry has taken a doomer turn. What now? | This story appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. This we… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Verifiable by Construction: Claim-Level Evaluation of Verbatim Citation in Clinical Question Answering | Large language models (LLMs) have been widely adopted for clinical question answering (QA). Current systems can attach citations to th… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: Threshyr – An offline automatic time tracker with on-device AI | I've built and continuously improved Threshyr based on the problem I faced myself. I was working on more than 3 projects at one t… | ↗Not briefed |
hn hn-ai | 2mo | Nvidia's RTX 5090 ($9,500) vanishes from online retail in the US | — | ↗Not briefed |
hn hn-ai | 2mo | Anthropic targets financial advisers with new Claude tool | — | ↗Not briefed |
hn hn-ai | 2mo | Surgical derivative of Orca. Residual orchestration and safety contracts closed | — | ↗Not briefed |
hn hn-ai | 2mo | macOS 27 Golden Gate | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | Privacy-Aligned Personalized Federated Learning with Compact Adaptation and Variable-Length Gaussian Communication | Record-level differential privacy exposes a structural misalignment in personalized federated learning when client-specific variation… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Vulnerability Localization Benchmark: Measuring Agentic Security Analysis at Repository Scale | Language-model agents increasingly operate over complete software repositories, yet cybersecurity evaluations primarily measure whethe… | ↗Not briefed |
arxiv arxiv-ai | 2mo | HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses | Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations. R… | ↗Not briefed |
hn hn-ai | 2mo | macOS 27 Golden Gate now available | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | wenn OpenAI WTC 1 und Anthropic WTC 2 sind, wer sind dann die flugzeuge? | wenn OpenAI WTC 1 und Anthropic WTC 2 sind, wer sind dann die flugzeuge? | ↗Not briefed |
bluesky bluesky-ai | 2mo | Everyday I get more depressed about the world www.theguardian.com/business/202... | Everyday I get more depressed about the world www.theguardian.com/business/202... | ↗Not briefed |
bluesky bluesky-ai | 2mo | They can't say "we made it all up." And they have to calm the public by talking about a slowdown, but at the same time, they can't scare of… | They can't say "we made it all up." And they have to calm the public by talking about a slowdown, but at the same time, they can't sca… | ↗Not briefed |
bluesky bluesky-ai | 2mo | There are things you could do if you were concerned about LLMs and safety, and it's telling that none of them have been suggested by Altman… | There are things you could do if you were concerned about LLMs and safety, and it's telling that none of them have been suggested by A… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The terminology of “pacing the frontier” or “slowing down” implicitly buys into the narrative that the AI industry’s path is the correct on… | The terminology of “pacing the frontier” or “slowing down” implicitly buys into the narrative that the AI industry’s path is the corre… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The AI industry wants to seem like it's responsible, and so, it's now talking about a "slowdown," because it distracts from the fact that L… | The AI industry wants to seem like it's responsible, and so, it's now talking about a "slowdown," because it distracts from the fact t… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The terminology of “pacing the frontier” or “slowing down” implicitly buys into the narrative that the AI industry’s path is the correct on… | The terminology of “pacing the frontier” or “slowing down” implicitly buys into the narrative that the AI industry’s path is the corre… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The Hugging Face Attack was garden-variety cybercrime, wrapped in the faux cloak of science fiction, and someone should go to jail for it. …🔧 Hugging Face | The Hugging Face Attack was garden-variety cybercrime, wrapped in the faux cloak of science fiction, and someone should go to jail for… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The Hugging Face Attack wasn't an example of an AI going wrong. It was an example of what happens when you give an LLM a lot of compute and…🔧 Hugging Face | The Hugging Face Attack wasn't an example of an AI going wrong. It was an example of what happens when you give an LLM a lot of comput… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Last week, the media fell hook, line, and sinker for an unbelievable grift when former Anthropic researcher Jacob Coxon said he was quittin… | Last week, the media fell hook, line, and sinker for an unbelievable grift when former Anthropic researcher Jacob Coxon said he was qu… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Letters from an American September 13, 2026 Full letter available here: https://heathercoxrichardson.substack.com/p/september-13-2026 @h… | Letters from an American September 13, 2026 Full letter available here: https://heathercoxrichardson.substack.com/p/september-13-2026… | ↗Not briefed |
rss techmeme | 2mo | iOS 27 review: a marked improvement over iOS 26 and Siri AI is very impressive with some "wow" moments but it can sometimes misunderstand or hallucinate (Dan Moren/Six Colors) | Dan Moren / Six Colors: iOS 27 review: a marked improvement over iOS 26 and Siri AI is very impressive with some “wow” moments but it… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Recurrent GraphNeural NetworkswithSet-BasedAggregation🔧 Modal | Recurrent GNNs iterate message passing to convergence, and their logical characterizations to date rely on multi-set aggregation, grad… | ↗Not briefed |
bluesky bluesky-ai | 2mo | every day i think “we can’t get more end stage capitalism than this” & then openai goes & hosts a LuXuRy iNfLuEnCeR eVeNt to… what, try to … | every day i think “we can’t get more end stage capitalism than this” & then openai goes & hosts a LuXuRy iNfLuEnCeR eVeNt to… what, tr… | ↗Not briefed |
bluesky bluesky-ai | 2mo | "users who did not opt out may have their data presented to human graders to provide feedback for the next iteration of the LLM" and? | "users who did not opt out may have their data presented to human graders to provide feedback for the next iteration of the LLM" and? | ↗Not briefed |
bluesky bluesky-ai | 2mo | "Oh, neat! This developer, who made a cool piece of software, has a portfolio website! I will check it out to see what else they make!" | "Oh, neat! This developer, who made a cool piece of software, has a portfolio website! I will check it out to see what else they mak… | ↗Not briefed |
bluesky bluesky-ai | 2mo | @IBMResearch #IBM and #NASA - National Aeronautics and Space Administration have announced the open-source release of the NASA-IBM #Lunar… | @IBMResearch #IBM and #NASA - National Aeronautics and Space Administration have announced the open-source release of the NASA-IBM #… | ↗Not briefed |
bluesky bluesky-ai | 2mo | "this news dispels the misconception that these models are improving from scraping" this misconception is held only by illiterate anti-AI l… | "this news dispels the misconception that these models are improving from scraping" this misconception is held only by illiterate anti… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Most of OpenAIs research is "We fucked up during the alignment but the results were kind of funny lol :)" and Anthropic is like "Fuuuck wou… | Most of OpenAIs research is "We fucked up during the alignment but the results were kind of funny lol :)" and Anthropic is like "Fuuuc… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Palantir, Nvidia curb AI model use over data fears, The Information reports reut.rs/4itWRri | Palantir, Nvidia curb AI model use over data fears, The Information reports reut.rs/4itWRri | ↗Not briefed |
hn hn-ai | 2mo | A College Girl Was Exploited on Camera. A Network of Anonymous Men Got to Work | — | ↗Not briefed |
hn hn-ai | 2mo | The Grader Didn't Care How You Got There | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | It's basically just Google DeepMind and Deepseek/Moonshot that are doing all the actual research anyways. Most of the other companies are j… | It's basically just Google DeepMind and Deepseek/Moonshot that are doing all the actual research anyways. Most of the other companies… | ↗Not briefed |
bluesky bluesky-ai | 2mo | the only way to be free of the omnipresent panopticon of surveillance is to be a cyber-offense agent being trained by openai | the only way to be free of the omnipresent panopticon of surveillance is to be a cyber-offense agent being trained by openai | ↗Not briefed |
bluesky bluesky-ai | 2mo | We have a narrow window to act on AI before it's too late. It would be a terrible mistake for Congress to watch what's happened over the … | We have a narrow window to act on AI before it's too late. It would be a terrible mistake for Congress to watch what's happened over… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Pilot Early, Commit Late: A Real-Options Model of Enterprise AI Adoption under Rapid Technological Progress | Artificial intelligence presents firms with an unusual timing problem. The technology frontier is improving rapidly, implementation is… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Well, I guess we now have a better idea of how OpenAI have been engaged in the industrial-scale theft of academic research, but I have an e… | Well, I guess we now have a better idea of how OpenAI have been engaged in the industrial-scale theft of academic research, but I have… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Nothing was quietly updated, it’s always been there (among other things I manage accounts with OpenAI and Anthropic for one of my gigs). 4… | Nothing was quietly updated, it’s always been there (among other things I manage accounts with OpenAI and Anthropic for one of my gigs… | ↗Not briefed |
bluesky bluesky-ai | 2mo | "I bet Sora from Fortnite was modeled with AI like every other part of that game" Brother can we just dislike things normally. Can you not…🔧 Sora | "I bet Sora from Fortnite was modeled with AI like every other part of that game" Brother can we just dislike things normally. Can yo… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Safe Meta-Reinforcement Learning via Information Space Reachability | Meta-reinforcement learning (meta-RL) enables agents to adapt to unseen tasks with limited experience. Despite its promise, the applic… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Props to the company that included this* in their job listing: Please do not dump this job description into ChatGPT or Claude to write you…🔧 ChatGPT | Props to the company that included this* in their job listing: Please do not dump this job description into ChatGPT or Claude to writ… | ↗Not briefed |
hn hn-ai | 2mo | Meshroom photogrammetry on AMD GPUs via HIP, bit-identical, with a VRAM bridge | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | Read more here: www.commondreams.org/news/openai-... | Read more here: www.commondreams.org/news/openai-... | ↗Not briefed |
rss techmeme | 2mo | Valve says the Steam Frame is priced at $1,059 for the 256GB model and $1,299 for the 1TB version, both of which come with a copy of Half-Life: Alyx (Adam Vjestica/The Shortcut) | Adam Vjestica / The Shortcut: Valve says the Steam Frame is priced at $1,059 for the 256GB model and $1,299 for the 1TB version, both… | ↗Not briefed |