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 |
|---|---|---|---|---|
arxiv arxiv-ai | 29d | Weighted Ensemble Sampling for Rare-Event Estimation in Language Models | Rare but severe failures of large language models can become practically significant at deployment scale. Estimating their probabiliti… | ↗Not briefed |
bluesky bluesky-ai | 29d | Govt agencies taking three months to report data breaches ia.acs.org.au/article/2026... | Govt agencies taking three months to report data breaches ia.acs.org.au/article/2026... | ↗Not briefed |
bluesky bluesky-ai | 29d | My maybe extremely hot take is that web design is visual art and making a website with a cloud-based LLM isn't functionally much different … | My maybe extremely hot take is that web design is visual art and making a website with a cloud-based LLM isn't functionally much diffe… | ↗Not briefed |
bluesky bluesky-ai | 29d | AI in your oven? Anthropic warns models ‘months away’ from self-replication on smart devices | AI in your oven? Anthropic warns models ‘months away’ from self-replication on smart devices | ↗Not briefed |
bluesky bluesky-ai | 29d | US indicts Saudi employee of OpenAI for allegedly bugging its offices ft.trib.al/s4XBV7H | US indicts Saudi employee of OpenAI for allegedly bugging its offices ft.trib.al/s4XBV7H | ↗Not briefed |
bluesky bluesky-ai | 29d | photo in here of the OpenAI engineer (allegedly!) checking on the listening device he secretly installed is an all-timer sfstandard.com/20… | photo in here of the OpenAI engineer (allegedly!) checking on the listening device he secretly installed is an all-timer sfstandard.c… | ↗Not briefed |
bluesky bluesky-ai | 29d | Holy shit. Can someone say techno futurist buddy movie where a Joe/Jane Average has to expose the damning secrets of a Epstein class AI cor… | Holy shit. Can someone say techno futurist buddy movie where a Joe/Jane Average has to expose the damning secrets of a Epstein class A… | ↗Not briefed |
rss techmeme | 29d | The FBI says it worked with CrowdStrike to dismantle 40+ laptop farms across Texas used to help North Koreans get remote US tech jobs under stolen identities (Dustin Volz/New York Times) | Dustin Volz / New York Times: The FBI says it worked with CrowdStrike to dismantle 40+ laptop farms across Texas used to help North Ko… | ↗Not briefed |
arxiv arxiv-ai | 29d | Embedding Geometry Shapes LLM Routing: Tracing Failure Modes and Improving Generalization | With the growing number of LLMs available, selecting which one to use for a prompt has become a key question. A large part of the rout… | ↗Not briefed |
bluesky bluesky-ai | 29d | Because my smart socket has enough power and storage to run an llm... | Because my smart socket has enough power and storage to run an llm... | ↗Not briefed |
bluesky bluesky-ai | 29d | Self-replication is neither a new development nor unique to genAI. The fact that a leading LLM company acts like it can't implement guardra… | Self-replication is neither a new development nor unique to genAI. The fact that a leading LLM company acts like it can't implement gu… | ↗Not briefed |
bluesky bluesky-ai | 29d | New, by me: An OpenAI employee on the product team has been arrested by the FBI and charged with planting a secret recording device in an O… | New, by me: An OpenAI employee on the product team has been arrested by the FBI and charged with planting a secret recording device in… | ↗Not briefed |
bluesky bluesky-ai | 29d | FBI arrests OpenAI employee accused of planting hidden recording device in conference room | FBI arrests OpenAI employee accused of planting hidden recording device in conference room | ↗Not briefed |
bluesky bluesky-ai | 29d | Saudi national OpenAI employee arrested after 'sophisticated recording device' planted in office | Saudi national OpenAI employee arrested after 'sophisticated recording device' planted in office | ↗Not briefed |
bluesky bluesky-ai | 29d | OpenAI hit with proposed class action over collection of children’s data reut.rs/4zwgd4i | OpenAI hit with proposed class action over collection of children’s data reut.rs/4zwgd4i | ↗Not briefed |
bluesky bluesky-ai | 29d | One could state, when if we're using full LLM/GenAI, "Well, simply inform the AI before the resources are wasted that this is an old painti… | One could state, when if we're using full LLM/GenAI, "Well, simply inform the AI before the resources are wasted that this is an old p… | ↗Not briefed |
bluesky bluesky-ai | 29d | I’m not convinced by this. You need to know quite a bit to put this stuff in place, even with an LLM to help. Maybe that will change but I’… | I’m not convinced by this. You need to know quite a bit to put this stuff in place, even with an LLM to help. Maybe that will change b… | ↗Not briefed |
bluesky bluesky-ai | 29d | The best regulation of AI would be to pass laws that hold the people in charge of companies like Open AI & Anthropic criminally responsible… | The best regulation of AI would be to pass laws that hold the people in charge of companies like Open AI & Anthropic criminally respon… | ↗Not briefed |
hn hn-ai | 29d | OpenAI employee arrested by FBI over hidden listening device, company says | — | ↗Not briefed |
bluesky bluesky-ai | 29d | It’s still possible to differentiate LLM generated text from human text, and there are interesting effects if you try to make it less detec… | It’s still possible to differentiate LLM generated text from human text, and there are interesting effects if you try to make it less… | ↗Not briefed |
bluesky bluesky-ai | 29d | Raffi Krikorian explains that while consumers flock to ChatGPT and Claude, companies are increasingly turning to open models that they can …🔧 ChatGPT | Raffi Krikorian explains that while consumers flock to ChatGPT and Claude, companies are increasingly turning to open models that they… | ↗Not briefed |
arxiv arxiv-ai | 29d | FireGraphNet: Sub-daily wildfire spread prediction with directed graph neural networks | Wildfire decisions are made in hours, yet most learned spread models predict daily on targets including burned area. We propose FireGr… | ↗Not briefed |
bluesky bluesky-ai | 29d | The biz model for AI startups is: 1. Burn lots of money upfront 2. Get consumers hooked on the product 3. Raise prices for consumers signi… | The biz model for AI startups is: 1. Burn lots of money upfront 2. Get consumers hooked on the product 3. Raise prices for consumers… | ↗Not briefed |
arxiv arxiv-ai | 29d | NEPSEE: A Double-Annotated Benchmark for Nepali Sentiment Analysis | Many sentiment datasets use one annotator for each item, so the test labels show the conventions of that person. We build a Nepali ben… | ↗Not briefed |
arxiv arxiv-ai | 29d | Adaptive Computation with Conformal Certification🔧 FLUX | Scientific imaging data involving medical (e.g., 3D radiographic, microscopic images) and astronomical sensors pose different challeng… | ↗Not briefed |
hn hn-ai | 29d | OpenAI engineer arrested in bugging of executive conference room | — | ↗Not briefed |
bluesky bluesky-ai | 29d | None of my smart home devices has enough compute or memory to load anything Anthropic makes. | None of my smart home devices has enough compute or memory to load anything Anthropic makes. | ↗Not briefed |
bluesky bluesky-ai | 29d | Not telling anyone what to enjoy, but assume any gee-whiz neat software project that seemed to pop up overnight was built with cloud-based … | Not telling anyone what to enjoy, but assume any gee-whiz neat software project that seemed to pop up overnight was built with cloud-b… | ↗Not briefed |
arxiv arxiv-ai | 29d | Identity, Incentives, and Cooperative Behavior for AI: Reverse-Engineering Multiagent Cooperation | The behavior of AI systems increasingly depends on the extent to which agents cooperate or coordinate with one another, in ways that m… | ↗Not briefed |
arxiv arxiv-ai | 29d | Structure-aware Bayesian full waveform inversion via diffusion-regularized Stein variational gradient descent | Full waveform inversion (FWI) is a powerful tool for high-resolution subsurface model reconstruction, but it remains highly nonlinear… | ↗Not briefed |
bluesky bluesky-ai | 29d | Thought I'd share this post from a friend which is a more nuanced version of the frustration among mathematicians about how the "mathocalyp… | Thought I'd share this post from a friend which is a more nuanced version of the frustration among mathematicians about how the "matho… | ↗Not briefed |
hn hn-ai | 29d | Commodore 64 Raycasting with variable ceiling and floor heights | — | ↗Not briefed |
bluesky bluesky-ai | 29d | “OpenAI described the device as sophisticated and designed to evade detection” it was wrapped in white toilet paper behind a black televisi… | “OpenAI described the device as sophisticated and designed to evade detection” it was wrapped in white toilet paper behind a black tel… | ↗Not briefed |
arxiv arxiv-ai | 29d | How to Evaluate When You Don't Know How to Evaluate? Choosing Reliable Setups for Base Language Model Benchmarking | Evaluation of large language models (LLMs) remains largely pre-scientific: it involves many design decisions, with little guidance on… | ↗Not briefed |
arxiv arxiv-ai | 29d | HERO: Harnessing Evidence for Test-time Refinement and Prompt Optimization in Program Generation | Harness optimization improves LLMs by refining reusable prompts and scaffolding, yet requires further adaptation for program generatio… | ↗Not briefed |
bluesky bluesky-ai | 29d | The recording device was apparently discovered plugged into a TV’s USB port in an executive meeting room. www.theverge.com/ai-artificia... | The recording device was apparently discovered plugged into a TV’s USB port in an executive meeting room. www.theverge.com/ai-artifici… | ↗Not briefed |
bluesky bluesky-ai | 29d | Better to just say 'Anthropic set a fire/Anthropic caused an oil spill, which then spread and hurt a lot of people' imho, in terms of both … | Better to just say 'Anthropic set a fire/Anthropic caused an oil spill, which then spread and hurt a lot of people' imho, in terms of… | ↗Not briefed |
bluesky bluesky-ai | 29d | If an AI person is claiming that Microsoft Excel isn't conscious, but the LLM they're selling is, then ... they should be able to point to … | If an AI person is claiming that Microsoft Excel isn't conscious, but the LLM they're selling is, then ... they should be able to poin… | ↗Not briefed |
hn hn-ai | 29d | The AI Curiosity Trap | — | ↗Not briefed |
bluesky bluesky-ai | 29d | been waiting outside the window in my sickos shirt for 3 years for this | been waiting outside the window in my sickos shirt for 3 years for this | ↗Not briefed |
rss theverge-ai | 29d | OpenAI employee arrested for allegedly planting a bug in a conference room | An OpenAI engineer was arrested after the company said it discovered a "sophisticated recording device" hidden in one of its offices.… | ↗Not briefed |
arxiv arxiv-ai | 29d | Learning in Strategic Queuing with Geometric Recharge: From Bandits to Systems | We study packet routing in queuing systems where queues act as selfish, myopic agents that use no-regret learning algorithms to select… | ↗Not briefed |
bluesky bluesky-ai | 29d | that thumbnail is what the anthropic guy looks like? he has the audacity to say an adult death is 2 or 3 times worse than an infant's death… | that thumbnail is what the anthropic guy looks like? he has the audacity to say an adult death is 2 or 3 times worse than an infant's… | ↗Not briefed |
arxiv arxiv-ai | 29d | Diagnosing Low-Confidence Patient Profiles via Localized Harmful Covariate Shift Detection | Before deploying clinical prediction models, their reliability must be assessed to ensure safe use. Errors may arise from distribution… | ↗Not briefed |
arxiv arxiv-ai | 29d | Multi-System Cognitive Steering for Generative Communication | A fundamental bottleneck for translating thoughts to speech is the high physical cost of motor input forcing a strict tradeoff between… | ↗Not briefed |
bluesky bluesky-ai | 29d | This is absolutely insane. Serious AI psychosis. "Its better at coding than engineers! I mean, its not better at coding when you take into… | This is absolutely insane. Serious AI psychosis. "Its better at coding than engineers! I mean, its not better at coding when you take… | ↗Not briefed |
bluesky bluesky-ai | 29d | Exactly. I can barely get an LLM running locally on my expensive GPU with 16GB RAM to do anything useful. As if these models are going to … | Exactly. I can barely get an LLM running locally on my expensive GPU with 16GB RAM to do anything useful. As if these models are goin… | ↗Not briefed |
arxiv arxiv-ai | 29d | Learning from Self-Generated Responses: Reliability--Uncertainty Self-Distillation for Reasoning | Reference solutions and question-specific rubrics are not always available to guide the teacher in self-distillation. A large language… | ↗Not briefed |
rss techmeme | 29d | Xbox CEO Asha Sharma announces Xbox Labs, a small team to test emerging ideas, citing recent game ports and mashups, many of them built with AI tools (Jon Hicks/GamesIndustry.biz) | Jon Hicks / GamesIndustry.biz: Xbox CEO Asha Sharma announces Xbox Labs, a small team to test emerging ideas, citing recent game ports… | ↗Not briefed |
bluesky bluesky-ai | 29d | Countdown to random SWATing by AI arstechnica.com/tech-policy/... | Countdown to random SWATing by AI arstechnica.com/tech-policy/... | ↗Not briefed |