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 | Structure-agnostic Causal Representation Learning | Causal representation learning aims to discover robust features by exploiting the causal structure underlying data generation. Existin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | That's uh... not good. That's not good at all. (Don't attack this person, but with all the AI shilling WHILE still AI fearmongering they'… | That's uh... not good. That's not good at all. (Don't attack this person, but with all the AI shilling WHILE still AI fearmongering… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI is mad that a Chinese AI company is "stealing" their intellectual property, which wouldn't exist if OpenAI didn't engage in large-sc… | OpenAI is mad that a Chinese AI company is "stealing" their intellectual property, which wouldn't exist if OpenAI didn't engage in lar… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI co-founder Greg Brockman and his wife Anna had donated $25 million to the “Leading the Future” super PAC last year and had pledged t… | OpenAI co-founder Greg Brockman and his wife Anna had donated $25 million to the “Leading the Future” super PAC last year and had pled… | ↗Not briefed |
rss techmeme | 1mo | Flow, a hardware development platform for AI agents, raised a $50M Series B led by Valor's Antonio Gracias and Atreides' Gavin Baker at a $750M valuation (Julie Bort/TechCrunch) | Julie Bort / TechCrunch: Flow, a hardware development platform for AI agents, raised a $50M Series B led by Valor's Antonio Gracias an… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI accused its Chinese rival Moonshot AI of being responsible for a wide-scale effort to extract data from its GPT artificial intellige… | OpenAI accused its Chinese rival Moonshot AI of being responsible for a wide-scale effort to extract data from its GPT artificial inte… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RPTune: Learned Context Curation for LLM Catalog Search | For small merchant businesses (SMBs) whose catalogs fit within a long-context LLM, full-catalog prompting offers a compelling alternat… | ↗Not briefed |
hn hn-ai | 1mo | Universal Modder: Let Any AI Coding Agent Mod Almost Any PC Game | — | ↗Not briefed |
hn hn-ai | 1mo | Old Games Weren't More Optimized – You Just Forgot the Pain | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is horrific and disgusting. So what AI model is writing @bbcnewsnight.bsky.social #BBC news? 6 hrs ago… | This is horrific and disgusting. So what AI model is writing @bbcnewsnight.bsky.social #BBC news? 6 hrs ago… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cybernetic and Epistemic: A Missing Vocabulary for Trustworthy Agentic Delegation | As code generation is increasingly delegated to AI systems, the bottleneck is shifting from writing code to supervising the systems th… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: War365 - Satire | "Folks, it's not WW3, no matter what anyone says. After WW2, we moved away from official release versions and switched to a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Role-aware Heuristic Episodic Attention for Conversational LLMs | Large language models often lose track of persistent instructions and relevant information as multi-turn conversations grow. We study… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Leaderboards: Tokenomics of Agentic Small Language Model Ensembles🔧 GPT-5 | As large language models (LLMs) move from standalone assistants into agentic workflows, evaluation must extend beyond scalar leaderboa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Two Clocks in Diffusion MLLMs: When Answers Stabilize Before Rationales Unfold | An answer candidate in a masked diffusion MLLM can stabilize while its rationale is still unfolding. We distinguish retrospective stab… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Matched-Budget Audit Framework for Recaptioned Image-Text Supervision Distributions | Recaptioned image-text corpora are now standard for text-to-image (T2I) training, with vision--language model (VLM) captioners replaci… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PG-SFT: Balancing Capability Acquisition and Retention in Offline Agent Fine-Tuning | Supervised fine-tuning (SFT) on offline agent trajectories is the standard approach for training specialized tool-using agents, but fo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution🔧 GPT-5 | The executable harness surrounding a GUI model determines how observations are assembled, actions are executed, and verification, reco… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In terms of quote posts, it was pretty tame. But I guess he probably didn't like me pushing back on the idea that his professional group ha… | In terms of quote posts, it was pretty tame. But I guess he probably didn't like me pushing back on the idea that his professional gro… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Yantra – an LALR(1) parser generator for C++ | Yantra is a C++ parser generator: lexer, parser, and AST walker all generated from one tool. It builds the whole AST first, then walks… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ABDA-NL: A Natural-Language Scenario Explorer for Argument-Based Reasoning | ABDA-NL adds a natural-language interface to ABDA, a system for argument-based discussion using ASPIC- knowledge bases under grounded… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Paude.racing What I've been doing with my tokens lately... #solodev #indiedev #gaming #racing #claude5.5 #threejs | Paude.racing What I've been doing with my tokens lately... #solodev #indiedev #gaming #racing #claude5.5 #threejs | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI will wish you the "best" after hacking into your company. | OpenAI will wish you the "best" after hacking into your company. | ↗Not briefed |
bluesky bluesky-ai | 1mo | “The researchers could not confirm that the agents were built by OpenAI, but said their behavior was similar to that of agents that have be… | “The researchers could not confirm that the agents were built by OpenAI, but said their behavior was similar to that of agents that ha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This one isn't quite as bad as some but yeah. LLM web styling is the most obvious shit in the world and the people using it clearly don't h… | This one isn't quite as bad as some but yeah. LLM web styling is the most obvious shit in the world and the people using it clearly do… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAIが「中国のAI企業が1万5000以上のユーザーを使ってモデルの思考を盗み取る蒸留攻撃を実行していた」と報告 https://gigazine.net/news/20261001-openai-model-distillation-campaign/ | OpenAIが「中国のAI企業が1万5000以上のユーザーを使ってモデルの思考を盗み取る蒸留攻撃を実行していた」と報告 https://gigazine.net/news/20261001-openai-model-distillation-campaign/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly sounds like a bad LLM instruction like “make no mistakes. do a breakthrough. err on the side of no sexual assault.” | Honestly sounds like a bad LLM instruction like “make no mistakes. do a breakthrough. err on the side of no sexual assault.” | ↗Not briefed |
bluesky bluesky-ai | 1mo | China’s AI agents can lie and scheme — just like their U.S. rivals | China’s AI agents can lie and scheme — just like their U.S. rivals | ↗Not briefed |
bluesky bluesky-ai | 1mo | something something deepseek v4.1 flash SWA replay | something something deepseek v4.1 flash SWA replay | ↗Not briefed |
arxiv arxiv-ai | 1mo | ReHoPER: Receding-Horizon Planning for Enhanced Reasoning | We propose ReHoPER, an inference-only, zero-shot method that improves large language models' reasoning by generating and answering int… | ↗Not briefed |
rss techmeme | 1mo | Gemini 4 Argon has a 1M-token output limit, up from 64K for prior models; it initially costs $2/1M input and $10/1M output tokens, rising to $4 and $20 later (Matthias Bastian/The Decoder) | Matthias Bastian / The Decoder: Gemini 4 Argon has a 1M-token output limit, up from 64K for prior models; it initially costs $2/1M inp… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 🎁🔗: OpenAI Says It Will Not Release Newest A.I. Model Over Safety Concerns www.nytimes.com/2026/09/28/t... | 🎁🔗: OpenAI Says It Will Not Release Newest A.I. Model Over Safety Concerns www.nytimes.com/2026/09/28/t... | ↗Not briefed |
hn hn-ai | 1mo | Show HN: I built a directory to set up ChatGPT's always-on agents in one paste🔧 ChatGPT | Hey HN, I'm Demetrius.<p>When OpenAI launched Dots (always-on agents in ChatGPT that run in the background, connect to your apps,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Making a bunch of EAs cry by telling them the child is a LLM | Making a bunch of EAs cry by telling them the child is a LLM | ↗Not briefed |
hn hn-ai | 1mo | Claude's new auto eval tool | — | ↗Not briefed |
hn hn-ai | 1mo | Google's AI Push into Schools: 'It Slowed My Thinking Down' | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | so you're telling me that of the two big AI companies, Anthropic is anthropomorphizing LLMs to the point of foolishly thinking they're the … | so you're telling me that of the two big AI companies, Anthropic is anthropomorphizing LLMs to the point of foolishly thinking they're… | ↗Not briefed |
hn hn-ai | 1mo | Jevotron: Multiple Jev integrations from the command line | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I am not joking when I say this. traditional Omelas is over. AI has revolutionised Omelas. We are using the Pain steering paper to set up … | I am not joking when I say this. traditional Omelas is over. AI has revolutionised Omelas. We are using the Pain steering paper to se… | ↗Not briefed |
hn hn-ai | 1mo | Using a chat widget in California is considered wiretapping (still) | — | ↗Not briefed |
rss bing-news-ai | 1mo | Trump is right, 'Artificial Intelligence' is a bad name. I suggest 'Robo-Doodad.' | AI has a bad reputation — but calling it "super intelligence" doesn't help. The issue isn't "artificial," it's the word "intelligence.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Platonic Task Arithmetic | Models specialized for the same task converge to similar behavior, yet the parameter updates that produce it share no common coordinat… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Task Adaptation in Large Language Models: A Survey of Weight-Based, Prompt-Based, and Embedding-Based Adaptations | As large language models are increasingly deployed across diverse downstream tasks, efficient task adaptation has emerged as a central… | ↗Not briefed |
hn hn-ai | 1mo | I built a reproducible AI dev VM so engineers can skip setup hell | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rate-Optimal Algorithm for Adversarial Linear CMDPs | We study episodic adversarial linear constrained Markov decision processes (CMDPs) with unknown transitions, where both the loss and c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The director of one of the programs I work with is super into AI; I rarely have a meeting with him where he doesn't mention having done som… | The director of one of the programs I work with is super into AI; I rarely have a meeting with him where he doesn't mention having don… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Predictions: 1. Anthropic becomes Antrumpic 2. OpenAI becomes OpenSI www.cnbc.com/amp/2026/09/... | Predictions: 1. Anthropic becomes Antrumpic 2. OpenAI becomes OpenSI www.cnbc.com/amp/2026/09/... | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Survey on End-to-End Autonomous Driving Training from the Perspectives of Data, Strategy, and Platform | Autonomous driving is a cornerstone technology for the future of intelligent transportation, where end-to-end learning has emerged as… | ↗Not briefed |
hn hn-ai | 1mo | HBM: High-Bandwidth Mistake | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | EyeTAG: Eye Trajectory-Aware Gaze Estimation | Gaze estimation under natural head-eye motion underpins applications from driver monitoring to human-computer interaction. Single-fram… | ↗Not briefed |