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 |
|---|---|---|---|---|
rss techmeme | 28d | Google plans to acquire power from AEP's Cheyenne Fuel Cell under a 20-year deal; AEP will invest ~$4B in the Wyoming fuel cell project using Bloom's tech (Kelly Cloonan/Wall Street Journal) | Kelly Cloonan / Wall Street Journal: Google plans to acquire power from AEP's Cheyenne Fuel Cell under a 20-year deal; AEP will invest… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Veduta – Personal Agent with GenUI | The idea came to me while I was watching a livestream with Antirez and mr RIP, one of the first streams they did together. I thought A… | ↗Not briefed |
bluesky bluesky-ai | 28d | SLOP WATCH: Forza Horizon 1 Reimagined is AI slop. All the hallmarks of Claude are here, including mile-long comments and commit notes. Th… | SLOP WATCH: Forza Horizon 1 Reimagined is AI slop. All the hallmarks of Claude are here, including mile-long comments and commit note… | ↗Not briefed |
hn hn-ai | 28d | Experimenting with Dev Containers and Permission Allowlists | — | ↗Not briefed |
bluesky bluesky-ai | 28d | Wow! #OpenAI has a soul, people. You should trust it! We will align our latest data collection surveillance bot so that you can't poison … | Wow! #OpenAI has a soul, people. You should trust it! We will align our latest data collection surveillance bot so that you can't po… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Approximatr – The daily game of spatial intuition | Approximatr uses your built-in ability to gauge sizes, distances, alignments, angles etc to see how good your eyes are!<p>I've be… | ↗Not briefed |
hn hn-ai | 28d | Just Using Claude No Longer Generates Alpha | — | ↗Not briefed |
hn hn-ai | 28d | Show HN: Markdown to Doc in a Click | Everything is markdown now thanks to AI agents, which is honestly quite convenient for all things development, but once in a while you… | ↗Not briefed |
bluesky bluesky-ai | 28d | anthropic so desperately wants to be Cyberpunk Rache Bartmoss and the DataKrash and ALSO want to be Netwatch to set up and profit from an i… | anthropic so desperately wants to be Cyberpunk Rache Bartmoss and the DataKrash and ALSO want to be Netwatch to set up and profit from… | ↗Not briefed |
bluesky bluesky-ai | 28d | This is an utter fabrication. In law, LLM mistakes are tolerated at a rate that would quickly get a human fired, and AI output is incapable… | This is an utter fabrication. In law, LLM mistakes are tolerated at a rate that would quickly get a human fired, and AI output is inca… | ↗Not briefed |
rss techmeme | 28d | Booking Holdings launches Vayo, a travel startup that uses AI agents to track booking confirmations, price drops, and more, with a planned $100M investment (Todd Bishop/GeekWire) | Todd Bishop / GeekWire: Booking Holdings launches Vayo, a travel startup that uses AI agents to track booking confirmations, price dro… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Codepill – New atomic licensing scheme | Proposal specification: <a href="https://x-ryl669.github.io/codepill/" rel="nofollow">https://x-ryl669.g… | ↗Not briefed |
arxiv arxiv-ai | 28d | Cross-Model In-Context Neurofeedback Reveals (Limited) Activation Control in LLMs | Probes on activation vectors are widely used to detect misbehavior in language models, but their success relies on the assumption that… | ↗Not briefed |
rss the-decoder | 28d | AI guardrails designed to stop hackers are now blocking defenders, and Orion-1 aims to fill the gap | Artemis Security's new Orion-1 model is purpose-built for cyber defense and beats Claude Opus 5.5 and GPT-6 Sol on the company's own b… | ↗Not briefed |
bluesky bluesky-ai | 28d | If you want to get a good laugh, point Google AI (search engine version, not Pro or Argon) at one of the OpenAI proofs not formalized in le… | If you want to get a good laugh, point Google AI (search engine version, not Pro or Argon) at one of the OpenAI proofs not formalized… | ↗Not briefed |
bluesky bluesky-ai | 28d | Lawsuit wants OpenAI equity awarded to minors who use ChatGPT and a declaration that the site's terms of use are void for minors🔧 ChatGPT | Lawsuit wants OpenAI equity awarded to minors who use ChatGPT and a declaration that the site's terms of use are void for minors | ↗Not briefed |
hn hn-ai | 28d | Gemini 4: A New Hope for AI Hackers | — | ↗Not briefed |
arxiv arxiv-ai | 28d | TreeSpark: Learning to Draft the Trees You Verify for Speculative Decoding | Parallel drafters such as DFlash and DSpark reduce proposal latency, but early mismatches limit how much of each block can be accepted… | ↗Not briefed |
arxiv arxiv-ai | 28d | RM-OPD: On-Policy Distillation for Generative Reward Models | Reinforcement Learning on Mixed-domain preference data (Mix-RL) has become the mainstream approach for training general-purpose Genera… | ↗Not briefed |
arxiv arxiv-ai | 28d | MultiPartyTurnBench: Benchmarking Floor Control in Multi-Party Spoken Dialogue | Full-duplex speech systems must decide when to enter, hold, or relinquish the conversational floor, even as multiple speakers overlap… | ↗Not briefed |
rss techmeme | 28d | Amazon Alexa Tablet 12 Pro review: gorgeous screen, Android apps, and nice design and build quality, but Amazon's bloatware is pervasive and Alexa+ is subpar (David Pierce/The Verge) | David Pierce / The Verge: Amazon Alexa Tablet 12 Pro review: gorgeous screen, Android apps, and nice design and build quality, but Ama… | ↗Not briefed |
arxiv arxiv-ai | 28d | Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming | We study the problem of learning accurate and cognitively simple binary classifiers that are directly interpretable as Boolean rules.… | ↗Not briefed |
bluesky bluesky-ai | 28d | i've got agents starting up agents starting up agents. i've burned a billion tokens before even getting to "hello world". anthropic legally… | i've got agents starting up agents starting up agents. i've burned a billion tokens before even getting to "hello world". anthropic le… | ↗Not briefed |
bluesky bluesky-ai | 28d | *even if* you justified the theft required to train AI LLM models, *even if* you're like "well everything else is bad for the environment t… | *even if* you justified the theft required to train AI LLM models, *even if* you're like "well everything else is bad for the environm… | ↗Not briefed |
arxiv arxiv-ai | 28d | Gap Detection: Mitigating Homogenization through Demand and Defensibility | Language models often produce semantically similar answers to open-ended questions, even when many plausible alternatives exist. Gener… | ↗Not briefed |
hn hn-ai | 28d | Show HN: An embeddable DOCX/XLSX/PPTX editor SDK for your AI agent | — | ↗Not briefed |
bluesky bluesky-ai | 28d | Then STOP! | Then STOP! | ↗Not briefed |
arxiv arxiv-ai | 28d | LLM-Based Policy Compliance Verification of Loan Applications | Banks rely on extensive internal policy frameworks to ensure that corporate lending decisions comply with regulatory requirements and… | ↗Not briefed |
bluesky bluesky-ai | 28d | No, please share this far and wide: the test that Anthropic is crowing about was done with a bunch of Android devices that were given a gia… | No, please share this far and wide: the test that Anthropic is crowing about was done with a bunch of Android devices that were given… | ↗Not briefed |
rss techmeme | 28d | Memo: Netflix is cutting 6% of its staff, or ~1,000 jobs globally across product, tech, legal, and HR, its biggest layoffs since 2022, amid slowing growth (Wall Street Journal) | Wall Street Journal: Memo: Netflix is cutting 6% of its staff, or ~1,000 jobs globally across product, tech, legal, and HR, its bigges… | ↗Not briefed |
bluesky bluesky-ai | 28d | Excited to share this work with Farhad Pashakhanloo, where we queried LLMs to derive the perceptual geometry of natural odor mixtures! We p… | Excited to share this work with Farhad Pashakhanloo, where we queried LLMs to derive the perceptual geometry of natural odor mixtures!… | ↗Not briefed |
bluesky bluesky-ai | 28d | There’s no such thing as an “ethical” LLM because when they say “ethically trained” they just mean the training data has been legally clear… | There’s no such thing as an “ethical” LLM because when they say “ethically trained” they just mean the training data has been legally… | ↗Not briefed |
rss techmeme | 28d | GlobalFoundries plans to build two "essential chip" plants in Malta, New York, backed by up to $21B from Japan; the US will take a 4% stake in GlobalFoundries (Bloomberg) | Bloomberg: GlobalFoundries plans to build two “essential chip” plants in Malta, New York, backed by up to $21B from Japan; the US will… | ↗Not briefed |
hn hn-ai | 28d | SpaceX activated 100mph speed-limit restriction on Starlink terminals in Ukraine | — | ↗Not briefed |
bluesky bluesky-ai | 28d | More talks about AI agents, sandboxes and investment funds stats from the panel "Safe doesn't mean slow" (speaking about frontier models)… | More talks about AI agents, sandboxes and investment funds stats from the panel "Safe doesn't mean slow" (speaking about frontier mo… | ↗Not briefed |
bluesky bluesky-ai | 28d | Anthropic Shows How Small AI Models Could Replicate Across IoT Devices Anthropic researchers show how small AI models could autonomously r…🔧 Replicate | Anthropic Shows How Small AI Models Could Replicate Across IoT Devices Anthropic researchers show how small AI models could autonomou… | ↗Not briefed |
bluesky bluesky-ai | 28d | As someone who has been a member of the Society of Authors for quite a while, I think I speak for many when I say that this subject should … | As someone who has been a member of the Society of Authors for quite a while, I think I speak for many when I say that this subject sh… | ↗Not briefed |
hn hn-ai | 28d | Google ships Gemini 4 Argon, Amex launches agent purchase authorization, plus | — | ↗Not briefed |
hn hn-ai | 28d | OWASP Hunting: An open-source security payload collection | — | ↗Not briefed |
bluesky bluesky-ai | 28d | AIの暴走、訴えるべき相手はだれなのか www.cnn.co.jp/tech/3525354... >百科事典サイト「ウィキペディア」を運営する非営利組織（NPO）、ウィキメディア財団は先週、米AI開発企業オープンAIのエージェントが無許可で一部のページに「悪意の疑いがある編集… | AIの暴走、訴えるべき相手はだれなのか www.cnn.co.jp/tech/3525354... >百科事典サイト「ウィキペディア」を運営する非営利組織（NPO）、ウィキメディア財団は先週、米AI開発企業オープンAIのエージェントが無許可で一部のページに「悪意の疑い… | ↗Not briefed |
bluesky bluesky-ai | 28d | I've contacted the SOA over the past few years asking for advice on starting a group action lawsuit against Anthropic and Open AI to get ju… | I've contacted the SOA over the past few years asking for advice on starting a group action lawsuit against Anthropic and Open AI to g… | ↗Not briefed |
arxiv arxiv-ai | 28d | Sparse Autoencoders for Mechanistic Interpretability in NLP: A Critical Survey of Recoverability, Interpretability, and Utility | Sparse autoencoders (SAEs) have emerged as a promising approach to mechanistic interpretability by decomposing language-model activati… | ↗Not briefed |
arxiv arxiv-ai | 28d | In-Context Batch Black-box Optimization via Flow Matching | Batch black-box optimization is important for expensive systems where many candidates can be evaluated in parallel, yet only a few rou… | ↗Not briefed |
arxiv arxiv-ai | 28d | Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms | Vision-language models (VLMs) are moving onto edge devices, but little is known about how much energy they draw on single-board comput… | ↗Not briefed |
rss arstechnica-tech | 28d | Anthropic warns small AI models could self-replicate through IoT devices🔧 Replicate | "Small weight-based replicators" could usher in a new generation of computer worms. | ↗Not briefed |
bluesky bluesky-ai | 28d | every day i look at job postings on linkedin and get a glimpse of the general feed on the way there, and every day i discover new kinds of … | every day i look at job postings on linkedin and get a glimpse of the general feed on the way there, and every day i discover new kind… | ↗Not briefed |
bluesky bluesky-ai | 28d | "We have seen over the past few months huge self-improvement of frontier AI and a series of incidents. Self regulation is not enough. We ar… | "We have seen over the past few months huge self-improvement of frontier AI and a series of incidents. Self regulation is not enough.… | ↗Not briefed |
hn hn-ai | 28d | Show HN: Train with Pasta. Workouts Powered by Apple Intelligence | Hi Hacker News! I'm Tim, and I made y'all some Pasta. :)<p>Others have New Year's resolutions to get fit, mine's t… | ↗Not briefed |
bluesky bluesky-ai | 28d | Not a surprise at all. Git had been and continues to be bad for mono repo development at large scale and large velocity of contributors. | Not a surprise at all. Git had been and continues to be bad for mono repo development at large scale and large velocity of contributor… | ↗Not briefed |
bluesky bluesky-ai | 28d | why do i have to click through this thread, scroll to the bottom, go to your website, go to a link attached to the website's page, scroll d… | why do i have to click through this thread, scroll to the bottom, go to your website, go to a link attached to the website's page, scr… | ↗Not briefed |