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 | 27d | You realize everything you think of as "AI" is owned by two companies, right? If you are pro-AI, you are explicitly in favor of letting Ope… | You realize everything you think of as "AI" is owned by two companies, right? If you are pro-AI, you are explicitly in favor of lettin… | ↗Not briefed |
hn hn-ai | 27d | Show HN: Explanations and follow-ons for OpenAI/math | The official openai/math repo is closed to issues, PRs, and discussion.<p>I made this app to share some explanatory videos that h… | ↗Not briefed |
bluesky bluesky-ai | 27d | Anthropic grants outside evaluator METR employee-level access in AI safety push reut.rs/47nWveV | Anthropic grants outside evaluator METR employee-level access in AI safety push reut.rs/47nWveV | ↗Not briefed |
bluesky bluesky-ai | 27d | "All of which raises the question: if someone running an operation so basic that — while working for an AI company — they were asking their… | "All of which raises the question: if someone running an operation so basic that — while working for an AI company — they were asking… | ↗Not briefed |
arxiv arxiv-ai | 27d | The Model Knows Better: Defending Medical Multi-Agent Systems Against Prompt Injection | Medical multi-agent systems, in which LLM agents discuss a case, are increasingly studied for clinical decision support. Prompt inject… | ↗Not briefed |
bluesky bluesky-ai | 27d | "Here’s Sam Altman again: “We see a future where intelligence is a utility, like electricity or water, and people buy it from us on a meter… | "Here’s Sam Altman again: “We see a future where intelligence is a utility, like electricity or water, and people buy it from us on a… | ↗Not briefed |
arxiv arxiv-ai | 27d | Flat Loss, Evolving Features: Grokking in Modular Addition | Can a network keep learning while its training cross-entropy stays nearly flat at a positive level? We show that it can. On modular ad… | ↗Not briefed |
bluesky bluesky-ai | 27d | If this is the case, then all it means is “conscious” is a meaningless concept for ethics and morality. An LLM does not and cannot have a… | If this is the case, then all it means is “conscious” is a meaningless concept for ethics and morality. An LLM does not and cannot h… | ↗Not briefed |
hn hn-ai | 27d | Space by Perplexity | — | ↗Not briefed |
bluesky bluesky-ai | 27d | New: Tech experts at OpenAI and Meta Platforms warned a federal judicial panel today that AI-generated "deepfakes" are becoming so sophisti… | New: Tech experts at OpenAI and Meta Platforms warned a federal judicial panel today that AI-generated "deepfakes" are becoming so sop… | ↗Not briefed |
bluesky bluesky-ai | 27d | LLM agents are conscious? these AI shills have a promising career in comedy when the bubble pops, looking at the goofy things they say | LLM agents are conscious? these AI shills have a promising career in comedy when the bubble pops, looking at the goofy things they say | ↗Not briefed |
bluesky bluesky-ai | 27d | AI is inevitable "Consulting giant Accenture has told many employees that the company will now only spend $29 a month per person on Claude… | AI is inevitable "Consulting giant Accenture has told many employees that the company will now only spend $29 a month per person on C… | ↗Not briefed |
arxiv arxiv-ai | 27d | SAG2: Embedding-Free Multi-Hop Retrieval with BM25 and JEV | Strong multi-hop retrieval should be practical to deploy without expensive corpus preparation or complex database infrastructure. We p… | ↗Not briefed |
arxiv arxiv-ai | 27d | Mitigating Demographic Disparities in Explanation-Human Alignment via Rationale Regularization | We study explanation fairness in hate speech detection. High classification performance does not always mean that a model produces equ… | ↗Not briefed |
hn hn-ai | 27d | Minions: A Virtual GPU Cloud Built on Idle Apple Devices | — | ↗Not briefed |
bluesky bluesky-ai | 27d | Alfozan, a 35 year-old Saudi national, OpenAI engineer who was fired after being charged this week with bugging a conference room was nabbe… | Alfozan, a 35 year-old Saudi national, OpenAI engineer who was fired after being charged this week with bugging a conference room was… | ↗Not briefed |
bluesky bluesky-ai | 27d | There's a story here that Accenture was paying Anthropic through the nose for their employees to dick around with Claude. And now they're… | There's a story here that Accenture was paying Anthropic through the nose for their employees to dick around with Claude. And now th… | ↗Not briefed |
arxiv arxiv-ai | 27d | Initiating Removal from Operational Symptoms in Federated Vehicular Networks | Connected vehicles train a driving model through federated learning (FL) and drive with it, so a faulty contribution often surfaces on… | ↗Not briefed |
arxiv arxiv-ai | 27d | Noise-Robust Forced Alignment for Spontaneous Speech | We show that forced aligners degrade on spontaneous, noisy real-world speech, and that fillers, laughter, and disfluencies corrupt wor… | ↗Not briefed |
bluesky bluesky-ai | 27d | The FTC is investigating possible consumer harms at two major AI companies. | The FTC is investigating possible consumer harms at two major AI companies. | ↗Not briefed |
bluesky bluesky-ai | 27d | In the last month, The Newsground has broken major stories on the arrested OpenAI employee Abdulrahman Alfozan, Serbs for Trump, Hunter Bid… | In the last month, The Newsground has broken major stories on the arrested OpenAI employee Abdulrahman Alfozan, Serbs for Trump, Hunte… | ↗Not briefed |
bluesky bluesky-ai | 27d | Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks ft.trib.al/SLV3WXO | Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks ft.trib.al/SLV3WXO | ↗Not briefed |
bluesky bluesky-ai | 27d | what's in for the tech companies..? they'll basically be obsolete since llm and the open source community can create whatever tools for fre… | what's in for the tech companies..? they'll basically be obsolete since llm and the open source community can create whatever tools fo… | ↗Not briefed |
bluesky bluesky-ai | 27d | Just the Docs: Eight state attorneys general file coordinated suits against OpenAI, alleging it concealed ChatGPT’s risks to public safety.…🔧 ChatGPT | Just the Docs: Eight state attorneys general file coordinated suits against OpenAI, alleging it concealed ChatGPT’s risks to public sa… | ↗Not briefed |
bluesky bluesky-ai | 27d | If it's so great soon developers will be training models on UE specifically and tailor our better, faster custom engines to avoid the 5% fe… | If it's so great soon developers will be training models on UE specifically and tailor our better, faster custom engines to avoid the… | ↗Not briefed |
bluesky bluesky-ai | 27d | As an example of why naively "fixing" stuff with an LLM without actually doing any investigation is a mistake: At work we had a weird infi… | As an example of why naively "fixing" stuff with an LLM without actually doing any investigation is a mistake: At work we had a weird… | ↗Not briefed |
bluesky bluesky-ai | 27d | “OpenAI alerta PF que tem gente planejando atentado contra Lula no ChatGPT” Eu não sei o que me choca mais: o extremismo de um atentado co…🔧 ChatGPT | “OpenAI alerta PF que tem gente planejando atentado contra Lula no ChatGPT” Eu não sei o que me choca mais: o extremismo de um atenta… | ↗Not briefed |
hn hn-ai | 27d | Show HN: ReliaMesh – Open-source reliability monitoring for AI agents | — | ↗Not briefed |
arxiv arxiv-ai | 27d | Embodiment-Classifier Guidance: Transferring Frozen VLAs Across Robot Dynamics without Demonstrations | Flow-based policies such as $π_0$ are trained on a fixed set of embodiments, and transferring them to new ones is an open problem. To… | ↗Not briefed |
bluesky bluesky-ai | 27d | Maternity leave during LLM discourse is pretty well-timed, ngl. Nice to have some distance as I work to keep a newborn alive | Maternity leave during LLM discourse is pretty well-timed, ngl. Nice to have some distance as I work to keep a newborn alive | ↗Not briefed |
bluesky bluesky-ai | 27d | Mehrere US-Bundesstaaten verklagen OpenAI #OpenAI #KünstlicheIntelligenz #ChatGPT🔧 ChatGPT | Mehrere US-Bundesstaaten verklagen OpenAI #OpenAI #KünstlicheIntelligenz #ChatGPT | ↗Not briefed |
bluesky bluesky-ai | 27d | I've only ever made sites as a hobbyist but it's really weird to see all the designs that look exactly the same, and my newest site is stru… | I've only ever made sites as a hobbyist but it's really weird to see all the designs that look exactly the same, and my newest site is… | ↗Not briefed |
bluesky bluesky-ai | 27d | If we part from the premise that LLM agents are conscious and then ask what's should we do about them, I think labour animals are the best … | If we part from the premise that LLM agents are conscious and then ask what's should we do about them, I think labour animals are the… | ↗Not briefed |
bluesky bluesky-ai | 27d | Update: Nick G. from Anthropic has responded on HN, and I believe that the attribution is now sufficient. I stand by my posts as of the tim… | Update: Nick G. from Anthropic has responded on HN, and I believe that the attribution is now sufficient. I stand by my posts as of th… | ↗Not briefed |
arxiv arxiv-ai | 27d | Structure-Aware Spatial Kernels for Gaussian Process Distributed Port-Hamiltonian Systems | Gaussian process distributed port-Hamiltonian systems (GP-dPHS) enable data-driven modeling of distributed dynamical systems while pre… | ↗Not briefed |
bluesky bluesky-ai | 27d | ❖ Oniichan dakedo Ai sae Areba Kankeinai yo ne! ┊ 98 sheets ❖ Model sheets (settei) for the 2012 TV series with character designs by Kosuk… | ❖ Oniichan dakedo Ai sae Areba Kankeinai yo ne! ┊ 98 sheets ❖ Model sheets (settei) for the 2012 TV series with character designs by… | ↗Not briefed |
bluesky bluesky-ai | 27d | There's absolutely going to be some small story in the next year about an employee unintentionally leaking schematics to a personal chatbot… | There's absolutely going to be some small story in the next year about an employee unintentionally leaking schematics to a personal ch… | ↗Not briefed |
bluesky bluesky-ai | 27d | We're not eviscerating AI techbros hard enough I hope killing a community with your Claude subscription was worth the karma on Reddit | We're not eviscerating AI techbros hard enough I hope killing a community with your Claude subscription was worth the karma on Reddit | ↗Not briefed |
hn hn-ai | 27d | TypeScript Types Declarations for WebMCP | — | ↗Not briefed |
bluesky bluesky-ai | 27d | Anthropic is letting METR researchers look under the hood with employee-level access | Anthropic is letting METR researchers look under the hood with employee-level access | ↗Not briefed |
rss techmeme | 27d | Sources: top DOJ officials halt career prosecutors' Biden-era investigation of Meta over facilitating illegal drug sales and are weighing more lenient options (Ben Penn/Bloomberg Law) | Ben Penn / Bloomberg Law: Sources: top DOJ officials halt career prosecutors' Biden-era investigation of Meta over facilitating illega… | ↗Not briefed |
bluesky bluesky-ai | 27d | I sort of think for a lot of purposes the right corporate AI front end locks you into a cheap model that only goes and gets successively be… | I sort of think for a lot of purposes the right corporate AI front end locks you into a cheap model that only goes and gets successive… | ↗Not briefed |
rss techcrunch-ai | 27d | Anthropic is letting METR researchers look under the hood with employee-level access | METR will publish its own assessments of Anthropic’s risks, incidents, and practices, and the frontier AI model maker has given up the… | ↗Not briefed |
arxiv arxiv-ai | 27d | Repairing LLM-Generated Specifications with Deterministic Diagnosis | Large language models (LLMs) are increasingly used as formalizers: given a reasoning problem expressed in natural language, they produ… | ↗Not briefed |
arxiv arxiv-ai | 27d | Who Do Perspectivist Models Predict Well? Annotator Agreement, Sparsity, and Performance | Successful completion of perspectivist natural language processing (NLP) annotation tasks requires algorithms to be aligned across div… | ↗Not briefed |
bluesky bluesky-ai | 27d | Spending $56k in LLM tokens to "decompile" a DS game with a source code already out in the wild. Why would you proudly announce this? | Spending $56k in LLM tokens to "decompile" a DS game with a source code already out in the wild. Why would you proudly announce this? | ↗Not briefed |
bluesky bluesky-ai | 27d | This. But also, maybe as a trans woman, it would be great to defend the idea of mankind first treating PEOPLE like they were PEOPLE (eg. t… | This. But also, maybe as a trans woman, it would be great to defend the idea of mankind first treating PEOPLE like they were PEOPLE (… | ↗Not briefed |
hn hn-ai | 27d | How to Use Jev with Claude Code, Codex and Cursor | — | ↗Not briefed |
hn hn-ai | 27d | Open-source self-hosted bot protection without CAPTCHAs | — | ↗Not briefed |
bluesky bluesky-ai | 27d | My kids have a lot of trouble with the fact that I won't let them use "AI" where that means ChatGPT, Gemini, etc, or like tablet games with…🔧 ChatGPT | My kids have a lot of trouble with the fact that I won't let them use "AI" where that means ChatGPT, Gemini, etc, or like tablet games… | ↗Not briefed |