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 | 1mo | DAGS: Disentangled Appearance-and-Geometry Steering of a Frozen Image DiT for Temporally Stabilized Generative Rendering | Diffusion transformers (DiTs) generate high-fidelity images from text and image conditions, but their outputs carry large variance and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic secretly invited religious leaders to SF to advise them on AI safety, had them sign NDAs, but then primarily tried to convince th… | Anthropic secretly invited religious leaders to SF to advise them on AI safety, had them sign NDAs, but then primarily tried to convin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | it's more important to help instill morality in the guys who run anthropic | it's more important to help instill morality in the guys who run anthropic | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Closure of Dynamical Systems with Kernel Ridge Regression | We develop a closure modeling framework for identifying missing components of dynamical systems using Kernel Ridge Regression (KRR). T… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OpenGameEval: Benchmarking Agentic Programming and Exploration in a Stateful Game Engine | We present OpenGameEval, a benchmark and evaluation framework for agentic game development inside Roblox Studio. It runs language mode… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: PhreshOS – OS for Web Apps | I'd like to share my experience building this system with you. It's an operating system for web applications, and it took me… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM model training costs are just going up and and up. Anthropic supposedly made a 4 BN dollar loss due to training costs being twice their… | LLM model training costs are just going up and and up. Anthropic supposedly made a 4 BN dollar loss due to training costs being twice… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Really cool to have a button I've used for years in an app to randomly change to one that will activate an LLM. Dudes I know your shit is t… | Really cool to have a button I've used for years in an app to randomly change to one that will activate an LLM. Dudes I know your shit… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neuron merging via inverse-activation regression for post-training compression of sigmoid neural networks | As neural networks continue to grow in scale, model compression is becoming increasingly important for efficient inference under limit… | ↗Not briefed |
hn hn-ai | 1mo | Claude-Shaped Science | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | DeepSeek recently rolled their Instant, Expert, and Visual models into one shared UI. You don't get to pick which one activates so it's be… | DeepSeek recently rolled their Instant, Expert, and Visual models into one shared UI. You don't get to pick which one activates so it… | ↗Not briefed |
bluesky bluesky-ai | 1mo | *briefly worried DeepSeek had fixed one of my hacks* Wait... *shows it a picture first* *works fine* I love how stupid AI companies are. | *briefly worried DeepSeek had fixed one of my hacks* Wait... *shows it a picture first* *works fine* I love how stupid AI companie… | ↗Not briefed |
arxiv arxiv-ai | 1mo | How to Have a Sensitive Debate: An Instance-Optimal Protocol for AI Debate | As powerful AI systems reach and sometimes surpass the abilities of human experts across a range of cognitively demanding tasks, the p… | ↗Not briefed |
rss techmeme | 1mo | New Mexico asks a judge to order Meta to pay up to $40B after a jury found it willfully deceived customers, in what could be the biggest US court penalty ever (KRQE) | KRQE: New Mexico asks a judge to order Meta to pay up to $40B after a jury found it willfully deceived customers, in what could be the… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Claude really shouldn't be left reminding academic writers to fill in their citations. What's going on here is scholarly delegation of basi… | Claude really shouldn't be left reminding academic writers to fill in their citations. What's going on here is scholarly delegation of… | ↗Not briefed |
hn hn-ai | 1mo | Opal: NPM-compatible package manager in Rust with crash-safe installs | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Test-time Multi-agent Coordination by Decomposed Value Gradient Flow🔧 Modal | Offline multi-agent reinforcement learning (MARL) faces a persistent trade-off. Expressive generative policies can represent multi-mod… | ↗Not briefed |
bluesky bluesky-ai | 1mo | WAIT EVERYONE! I have an idea! What if there were people who systematically studied consciousness that could lend wisdom into this situati… | WAIT EVERYONE! I have an idea! What if there were people who systematically studied consciousness that could lend wisdom into this si… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Out of Sync, Out of Sight: Phantom State Attacks against IIoT Intrusion Detection | Machine learning-based intrusion detection systems (IDS) are critical for securing Industrial Internet of Things (IIoT) environments.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evaluating Multi-Dimensional Generalization of Large Language Models in Temporal Extraction Tasks | Time and event expression extraction are fundamental temporal reasoning tasks, but the problem remains difficult due to annotation amb… | ↗Not briefed |
rss techmeme | 1mo | The US sanctions Russia-backed payment network A7, freezing its US assets and accusing it of enabling Iran and its proxy groups to evade western sanctions (Amy Mackinnon/Financial Times) | Amy Mackinnon / Financial Times: The US sanctions Russia-backed payment network A7, freezing its US assets and accusing it of enabling… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI has a LOT of work to do if they think Luna can compete with Jev | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reward Inflation: A Healthy Stimulus for Reinforcement Learning | Reward serves as the primary learning signal in reinforcement learning (RL). However, while reward magnitudes are typically held fixed… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI alerts more than 100 groups about rogue AI agent activity reut.rs/471akQi | OpenAI alerts more than 100 groups about rogue AI agent activity reut.rs/471akQi | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Sam Altman says we need to slow down development to ensure the safety of humanity," one campaigner said of OpenAI's CEO. "Yet he is allege… | "Sam Altman says we need to slow down development to ensure the safety of humanity," one campaigner said of OpenAI's CEO. "Yet he is a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling | World models (WMs) simulate the transition dynamics of environments, enabling agents to plan over the consequences of their actions. I… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ironic considering these things are formed by data collected by reading everything. Or maybe it really is from a person who just doesn't kn… | Ironic considering these things are formed by data collected by reading everything. Or maybe it really is from a person who just doesn… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Well this is odd. I got mail from someone or something at Anthropic (no way to tell which) inviting me to try out their new model. Um, no. … | Well this is odd. I got mail from someone or something at Anthropic (no way to tell which) inviting me to try out their new model. Um,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | For those without an AFR subscription, South Australian premier Peter Malinauskas has put a tech sector lobbyist and former policy liaison … | For those without an AFR subscription, South Australian premier Peter Malinauskas has put a tech sector lobbyist and former policy lia… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If AI does the work of 40 hours in 30, OpenAI says you should keep the 10 hours for yourself. They are officially backing the 4-day work we… | If AI does the work of 40 hours in 30, OpenAI says you should keep the 10 hours for yourself. They are officially backing the 4-day wo… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial Intelligence | As citizens embrace the technology for entertainment, relationships and even farming advice, the government is stepping up efforts to… | ↗Not briefed |
hn hn-ai | 1mo | Court Agrees with EFF: Utah's VPN Law Demands a Technical Impossibility | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | FAÇAM O FILME DA ANTHROPIC!!! JEREMY STRONG WE’RE GONNA GET YOU THAT OSCAR | FAÇAM O FILME DA ANTHROPIC!!! JEREMY STRONG WE’RE GONNA GET YOU THAT OSCAR | ↗Not briefed |
hn hn-ai | 1mo | Challengers Approach: Third Party App Stores Arrive to Google Play | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I mean, they're working on embodiments where the LLM is constantly churning! | I mean, they're working on embodiments where the LLM is constantly churning! | ↗Not briefed |
hn hn-ai | 1mo | Broadcom to lend Anthropic up to $42B to lease its chips, filing says | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | ENCORE: Exact Non-equilibrium COntrol with Replica Exchange for Diffusion Generation | Inference-time control steers a pretrained generative model towards a target distribution without retraining. We study tilted targets… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This gets into the "LLMs training on LLM-produced content" issue, which some people say will lead to recursive self-improvement but seems l… | This gets into the "LLMs training on LLM-produced content" issue, which some people say will lead to recursive self-improvement but se… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ironically, only an LLM can possibly consume every piece of media with artistic value, because only an LLM can process information fast eno… | Ironically, only an LLM can possibly consume every piece of media with artistic value, because only an LLM can process information fas… | ↗Not briefed |
hn hn-ai | 1mo | Pi v1.0.0 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic feeling a little left out today | Anthropic feeling a little left out today | ↗Not briefed |
rss wired-ai | 1mo | Whatever AI Safety Is, It’s Not This | Asking AI companies to self-regulate is a great way to pretend like you’ve accomplished something. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic pushes for opt-out model for Australian content as ABC warns of ‘cannibalisation’ of news www.theguardian.com/technology/2... | Anthropic pushes for opt-out model for Australian content as ABC warns of ‘cannibalisation’ of news www.theguardian.com/technology/2... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Pivot to AI is more profitable than OpenAI and Anthropic together please gimme $100b in my IPO it's AI | Pivot to AI is more profitable than OpenAI and Anthropic together please gimme $100b in my IPO it's AI | ↗Not briefed |
bluesky bluesky-ai | 1mo | The sheer cybersecurity threat of a brute force llm attack vs leaky llm programming | The sheer cybersecurity threat of a brute force llm attack vs leaky llm programming | ↗Not briefed |
bluesky bluesky-ai | 1mo | Just use a regular LLM like the rest of us lmao | Just use a regular LLM like the rest of us lmao | ↗Not briefed |
bluesky bluesky-ai | 1mo | I like to say that is was clearly the best of everything that came out of RR, which spoke very little of its qualities and most of the surr… | I like to say that is was clearly the best of everything that came out of RR, which spoke very little of its qualities and most of the… | ↗Not briefed |
hn hn-ai | 1mo | Pi-autoresearch ported to Claude Code 1:1 using the new mods API | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | これはよく思う〜。何パターンかサジェストまでしてくれるだけでいいし、処理負荷もLLM側は軽いはず（lunaやhaiku未満でも十分動きそう）だし、社会的便益が大きい | これはよく思う〜。何パターンかサジェストまでしてくれるだけでいいし、処理負荷もLLM側は軽いはず（lunaやhaiku未満でも十分動きそう）だし、社会的便益が大きい | ↗Not briefed |
hn hn-ai | 1mo | Scan your LLM agent's tools for cost-amplification bugs (no API key) | — | ↗Not briefed |