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 | 2d | ICYMI: Yesterday, we announced a $2.035 billion lawsuit against OpenAI. ChatGPT harms kids, and we’re demanding accountability. Watch the…🔧 ChatGPT | ICYMI: Yesterday, we announced a $2.035 billion lawsuit against OpenAI. ChatGPT harms kids, and we’re demanding accountability. Watc… | ↗Not briefed |
bluesky bluesky-ai | 2d | Imagine the market crash due to GTA and the Anthropic IPO in the same day? Would nature heal? | Imagine the market crash due to GTA and the Anthropic IPO in the same day? Would nature heal? | ↗Not briefed |
bluesky bluesky-ai | 2d | OMG you guys I just got an email from Amazon Studios asking about film/TV rights to Ancillary Justice! This is amazing! I am so excited! (n… | OMG you guys I just got an email from Amazon Studios asking about film/TV rights to Ancillary Justice! This is amazing! I am so excite… | ↗Not briefed |
arxiv arxiv-ai | 2d | Surrogate Modeling and Reduced-Order Methods in Gravitational-Wave Physics and Numerical Relativity: From Spectral Representations to Adaptive Compression | The rapid expansion of gravitational-wave astronomy and high-precision numerical relativity has created an acute computational bottlen… | ↗Not briefed |
bluesky bluesky-ai | 2d | Well shit. Looks like the LLM-generated copycat book problem from Amazon has infected Bookshop dot org. Went looking for Cobalt Red by Sidd… | Well shit. Looks like the LLM-generated copycat book problem from Amazon has infected Bookshop dot org. Went looking for Cobalt Red by… | ↗Not briefed |
bluesky bluesky-ai | 2d | The latest from my colleague @ucs.org, Dr Sara Al-Sayed, on potential paths toward multilateral AI risk-reduction and the benefits and limi… | The latest from my colleague @ucs.org, Dr Sara Al-Sayed, on potential paths toward multilateral AI risk-reduction and the benefits and… | ↗Not briefed |
hn hn-ai | 2d | "Anthropic Is Not a Safe AI Company" – Coalition Challenges Anthropic Safety | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: Can an AI Agent Bypass Your Website Security?🔧 Hugging Face | As part of our red-teaming efforts, we recently configured GPT-6 Astra to test how easily an AI agent can bypass existing bot detectio… | ↗Not briefed |
arxiv arxiv-ai | 2d | Induction Head Origins of SLAM-ASR Speech Recognition Models | What makes a small trained map from encoder states to the embedding space of a frozen language model sufficient for competitive speech… | ↗Not briefed |
arxiv arxiv-ai | 2d | DRIFT: Detachable Reasoning for Runtime-Flexible and Efficient End-to-End Driving | The integration of vision-language models (VLMs) into end-to-end (E2E) autonomous driving has demonstrated strong potential by providi… | ↗Not briefed |
bluesky bluesky-ai | 2d | Anthropic IPO Puts Wall Street’s Speculation Factory on Display www.bloomberg.com/news/article... | Anthropic IPO Puts Wall Street’s Speculation Factory on Display www.bloomberg.com/news/article... | ↗Not briefed |
arxiv arxiv-ai | 2d | Meta-LSQ: A Closed-Form Sobolev Inner Solver for Few-Shot Meta-Learning | Gradient-based meta-learners such as model-agnostic meta-learning (MAML) adapt to a new task with an inner loop of gradient descent th… | ↗Not briefed |
bluesky bluesky-ai | 2d | In a higher temperature setting, the LLM will choose another word sometimes So it might be | In a higher temperature setting, the LLM will choose another word sometimes So it might be | ↗Not briefed |
bluesky bluesky-ai | 2d | In order to stay coherent, the LLM needs to choose the most likely next word MOST of the time But in order to provide variable outputs, i… | In order to stay coherent, the LLM needs to choose the most likely next word MOST of the time But in order to provide variable outpu… | ↗Not briefed |
hn hn-ai | 2d | Microsoft rejected Rockstar's pitch to put GTA3 on the original Xbox (2021) | — | ↗Not briefed |
arxiv arxiv-ai | 2d | Openhagen Worlds: Calibrated Environments for Measuring and Training Scientific Discovery Agents | Agents that do science must choose informative experiments and reason from their results, but most discovery benchmarks cannot separat… | ↗Not briefed |
hn hn-ai | 2d | Dealing with LLMs in IETF Discussions | — | ↗Not briefed |
bluesky bluesky-ai | 2d | LOL so dumb | LOL so dumb | ↗Not briefed |
bluesky bluesky-ai | 2d | LLMs work via something called “next token prediction” Tokens are words, parts of words, sometimes two words The LLM ingests a large am… | LLMs work via something called “next token prediction” Tokens are words, parts of words, sometimes two words The LLM ingests a lar… | ↗Not briefed |
rss wired-ai | 2d | Criminals Have a New Favorite ‘Point-and-Shoot’ AI Hacking Tool | Artex and similar penetration testing tools allow newbie hackers to scan targeted systems for vulnerabilities and successfully wage at… | ↗Not briefed |
hn hn-ai | 2d | High school disables AI detector after it kept mistaking everyday items for guns | — | ↗Not briefed |
hn hn-ai | 2d | SetShot macOS app – snapshot and diff macOS settings changes | — | ↗Not briefed |
bluesky bluesky-ai | 2d | FWIW I agree. Vigorously! 🙃 I wasn’t offended and didn’t feel attacked! I just wanted to get it on the record because I also feel very str… | FWIW I agree. Vigorously! 🙃 I wasn’t offended and didn’t feel attacked! I just wanted to get it on the record because I also feel ver… | ↗Not briefed |
arxiv arxiv-ai | 2d | F$^3$: What Makes Good Experience in World-Action Models? | Deployed world-action models (WAMs) can keep improving only if they make good use of their own experience. Existing methods supply ret… | ↗Not briefed |
hn hn-ai | 2d | Ask HN: How do you reach the flow with vibe coding? | When I used to code without AI (probably not even one year ago), I sometimes reached a state of "flow" and it truly was a gr… | ↗Not briefed |
bluesky bluesky-ai | 2d | imo having the llm generated essay in front of it diminishes the notion of, in its words, "a community that values quality, performance, an… | imo having the llm generated essay in front of it diminishes the notion of, in its words, "a community that values quality, performanc… | ↗Not briefed |
bluesky bluesky-ai | 2d | Emphasizing this because folks noticed what appears to be LLM-generated prose (Pangram says 100%) in the Foreword to The Rust Programming L… | Emphasizing this because folks noticed what appears to be LLM-generated prose (Pangram says 100%) in the Foreword to The Rust Programm… | ↗Not briefed |
bluesky bluesky-ai | 2d | i’m as troubled by the weird formatting in the text box on the left. LLM generated as well, maybe | i’m as troubled by the weird formatting in the text box on the left. LLM generated as well, maybe | ↗Not briefed |
bluesky bluesky-ai | 2d | so wait if i run out of things to do at work can i convince them to give me openai cyber or claude mythos so i can use all of my domain kno… | so wait if i run out of things to do at work can i convince them to give me openai cyber or claude mythos so i can use all of my domai… | ↗Not briefed |
arxiv arxiv-ai | 2d | When Does Continual Pre-Training Help? A Controlled Study of Dialectal Adaptation for Québec French | Continual pre-training (CPT) is widely used to adapt pretrained language models to dialectal and non-standard language varieties, yet… | ↗Not briefed |
bluesky bluesky-ai | 2d | Yeah, I just mentioned this on a different branch of this thread, but want to reiterate here: not one word of the main text of the book had… | Yeah, I just mentioned this on a different branch of this thread, but want to reiterate here: not one word of the main text of the boo… | ↗Not briefed |
hn hn-ai | 2d | Ask HN: How do you use AI to get more relevant content? | The "algorithm" tries to give your attention something to eat. Sometimes, it's exactly what you want/need. Sometim… | ↗Not briefed |
arxiv arxiv-ai | 2d | Depth Before Breadth: Reverse-to-Forward Curriculum with Advantage Decomposition for LLM Agent | Reinforcement learning (RL) has become a common approach for training large language model (LLM) agents on long-horizon tasks. However… | ↗Not briefed |
rss techmeme | 2d | Cybersecurity company Xint says it found RCE vulnerabilities in Nginx using GPT 6.1 Sol and GPT-6 Luna; Wiz says it had separately found the same flaws (Thomas Brewster/Forbes) | Thomas Brewster / Forbes: Cybersecurity company Xint says it found RCE vulnerabilities in Nginx using GPT 6.1 Sol and GPT-6 Luna; Wiz… | ↗Not briefed |
bluesky bluesky-ai | 2d | A string of several posts from the Foundation and one from the Project that were clearly using LLMs to author text. For the record: not on… | A string of several posts from the Foundation and one from the Project that were clearly using LLMs to author text. For the record: n… | ↗Not briefed |
hn hn-ai | 2d | Show HN: RPC-dart – a bridge for tRPC/oRPC type safety for Flutter | There's no official support for tRPC/oRPC support in Flutter and using Node.JS with these libraries start to lose there bigg… | ↗Not briefed |
rss techmeme | 2d | Variational, which offers blockchain-based derivatives tied to cryptocurrencies and traditional assets, raised a $110M Series B at a $1.5B valuation (Jason Shubnell/The Block) | Jason Shubnell / The Block: Variational, which offers blockchain-based derivatives tied to cryptocurrencies and traditional assets, ra… | ↗Not briefed |
hn hn-ai | 2d | Show HN: Brrrty – Eleventy Ported to Rust | — | ↗Not briefed |
bluesky bluesky-ai | 2d | Is S Bill 5653 BAAJA, aka the Dirty Data Center bill, on your radar? It’s a dangerous anti-democratic bill that @schumer.senate.gov is help… | Is S Bill 5653 BAAJA, aka the Dirty Data Center bill, on your radar? It’s a dangerous anti-democratic bill that @schumer.senate.gov is… | ↗Not briefed |
arxiv arxiv-ai | 2d | Stealing Exact Training Data from Other Clients during Federated Learning | Malicious adversaries can employ gradient inversion attacks to reverse-engineer private data using the model updates of a round in Fed… | ↗Not briefed |
bluesky bluesky-ai | 2d | OpenAI CEO Sam Altman says it’s ‘hard to imagine the AI industry without Jensen’ | OpenAI CEO Sam Altman says it’s ‘hard to imagine the AI industry without Jensen’ | ↗Not briefed |
hn hn-ai | 2d | Project Gryffindor: Meta's plan to get more school kids using Instagram | — | ↗Not briefed |
hn hn-ai | 2d | Dissecting the GPU Load/Store Unit and Exposing Its Security Risks | — | ↗Not briefed |
bluesky bluesky-ai | 2d | Anthropic may emerge as one of the world’s largest companies and help renew investment in the AI industry during its public offering—but th… | Anthropic may emerge as one of the world’s largest companies and help renew investment in the AI industry during its public offering—b… | ↗Not briefed |
bluesky bluesky-ai | 2d | Doing a little more testing with this on prose and wow, Carbon really is a step change over Claude/GPT models. Still loves the emdash but t… | Doing a little more testing with this on prose and wow, Carbon really is a step change over Claude/GPT models. Still loves the emdash… | ↗Not briefed |
rss techcrunch-ai | 2d | OpenAI CEO Sam Altman says it’s ‘hard to imagine the AI industry without Jensen’ | Nvidia founder and former CEO Jensen Huang stepped down last week amid several scandals, including an affair with an alleged Chinese s… | ↗Not briefed |
bluesky bluesky-ai | 2d | I mean this is really it. They don't want you to say it, but this is it. Everything an LLM can do, a human can do. You just don't want to d… | I mean this is really it. They don't want you to say it, but this is it. Everything an LLM can do, a human can do. You just don't want… | ↗Not briefed |
bluesky bluesky-ai | 2d | 2027: Meet Claude Halibut, the Halibut-based AI agent for all your marine needs | 2027: Meet Claude Halibut, the Halibut-based AI agent for all your marine needs | ↗Not briefed |
hn hn-ai | 2d | Nobody Is Ready for This Future. But Nobody Is Stopping It | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: Jekyll-panel-TUI: TUI-style jekyll theme | — | ↗Not briefed |