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 | 1mo | I'm live now excited to play more BG3 with @nyyxveil.bsky.social !! I'll also be showing off a new png model! twitch.tv/liatriskitty | I'm live now excited to play more BG3 with @nyyxveil.bsky.social !! I'll also be showing off a new png model! twitch.tv/liatriskitty | ↗Not briefed |
arxiv arxiv-ai | 1mo | Population loss in shallow ReLU networks: Bias & families of critical points | The main result presented is a formula for the population loss in the student-teacher kernel model that is applicable to shallow ReLU… | ↗Not briefed |
hn hn-ai | 1mo | VeriTile: Formal Verification for Triton Kernels | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | DiffusionShadow: Diffusion-based Shadow Caching for Neural Volume Rendering | Implicit neural representations (INRs) have gained momentum in scientific visualization due to their compactness and scalability to la… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Prompt Injection Detection for Email Agents Through Attack Chain Modeling | Large language model email assistants are particularly vulnerable to indirect prompt injection because untrusted email content can be… | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.democracynow.org/2026/9/24/he... | www.democracynow.org/2026/9/24/he... | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.politico.com/news/2026/09... The White House has asked OpenAI and Anthropic not to share their new artificial intelligence models with … | www.politico.com/news/2026/09... The White House has asked OpenAI and Anthropic not to share their new artificial intelligence models… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Why doesn’t anyone anthropomorphize the engineers who are creating the AI models? For example: “OpenAI engineers wrote a software program…🔧 Hugging Face | Why doesn’t anyone anthropomorphize the engineers who are creating the AI models? For example: “OpenAI engineers wrote a software pr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Recursive Self-Improvement via On-Policy Distillation for Reasoning | On-policy distillation (OPD) trains a student model by having it generate trajectories, then matching its next-token predictions with… | ↗Not briefed |
hn hn-ai | 1mo | Paul Graham on LLMs 'Thinking' | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Causal Retention in Interactive Agents: Interface Factorization and Selective Adaptation | Task performance need not determine which intervention mechanism an agent retains. We study causal retention: whether a frozen learned… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Akamai signs $11.6 billion cloud deal with Anthropic, grants warrant for up to 5% stake reut.rs/4xOwqAz | Akamai signs $11.6 billion cloud deal with Anthropic, grants warrant for up to 5% stake reut.rs/4xOwqAz | ↗Not briefed |
hn hn-ai | 1mo | ServingStudio: Simulating, Analyzing, and Optimizing LLM Serving Systems | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic comes out every few days and is like “Our system accidentally stole all of the classified government files from England” but it n… | Anthropic comes out every few days and is like “Our system accidentally stole all of the classified government files from England” but… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Maybe he is using an llm to write for him and it might be more likely to use those sorts of neutral-to-positive terms, even if it's told to…🔧 ChatGPT | Maybe he is using an llm to write for him and it might be more likely to use those sorts of neutral-to-positive terms, even if it's to… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Re: ppl who are certain OpenAI was targeting health data for malicious reasons: it’s capitalism and they were looking for POC of applicatio… | Re: ppl who are certain OpenAI was targeting health data for malicious reasons: it’s capitalism and they were looking for POC of appli… | ↗Not briefed |
bluesky bluesky-ai | 1mo | frameworks for agents, coding tools you use, research agents, browser automation, voice agents and harnesses. also useful boring stuff: … | frameworks for agents, coding tools you use, research agents, browser automation, voice agents and harnesses. also useful boring st… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "OpenAI chose to email a generic inbox of Services Australia as its notification method. While OpenAI has not broken any laws in its notifi… | "OpenAI chose to email a generic inbox of Services Australia as its notification method. While OpenAI has not broken any laws in its n… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Claude, ai generate my full cock and balls ontop of this promoted vrc world. | Claude, ai generate my full cock and balls ontop of this promoted vrc world. | ↗Not briefed |
arxiv arxiv-ai | 1mo | MARCEDES: Score-based causal discovery under non-Gaussianity with continuous optimization | We consider the problem of learning the underlying causal directed acyclic graph (DAG) structure corresponding to a structural equatio… | ↗Not briefed |
bluesky bluesky-ai | 1mo | But my read this is ... barely... hacking as I would define it, nor is it a sign of superhuman competence with AI. If I told an informatio… | But my read this is ... barely... hacking as I would define it, nor is it a sign of superhuman competence with AI. If I told an infor… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI attacked Australia's health system and doubled down. Time to act | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Deciding that an LLM can think because the output appears thoughtful is like deciding to eat a toxic mushroom because it looks the same as … | Deciding that an LLM can think because the output appears thoughtful is like deciding to eat a toxic mushroom because it looks the sam… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Framework for Identifying, Categorizing, and Explaining Bias in AI-Generated Code | As Large Language Models (LLMs) become integrated into software development workflows, concerns regarding unintentional biases in AI-g… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think you're right, that's my mistake. They were using ExploitGym for evaluation, not clearly for training. Hacking-adjacent tasks like p… | I think you're right, that's my mistake. They were using ExploitGym for evaluation, not clearly for training. Hacking-adjacent tasks l… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI to preview GPT-6 Cyber within days, Fortune reports reut.rs/4AtwGrf | OpenAI to preview GPT-6 Cyber within days, Fortune reports reut.rs/4AtwGrf | ↗Not briefed |
bluesky bluesky-ai | 1mo | A friend revealed to me that his essay was written with Claude AI. “Yeah, I could tell,” I said. “But you’re a real writer,” he said, soun… | A friend revealed to me that his essay was written with Claude AI. “Yeah, I could tell,” I said. “But you’re a real writer,” he said,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this whole OpenAI Australia hack/breach story is very disappointing. If AI is going to hack into our government, i'd hope it would at least… | this whole OpenAI Australia hack/breach story is very disappointing. If AI is going to hack into our government, i'd hope it would at… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Kkkkk @oglobo.globo.com vergonhosamente dizendo q a anthropic apresentou um caso de "descoberta biológica" feita pela AI mdssss | Kkkkk @oglobo.globo.com vergonhosamente dizendo q a anthropic apresentou um caso de "descoberta biológica" feita pela AI mdssss | ↗Not briefed |
bluesky bluesky-ai | 1mo | 📦 NVIDIA / Model-Optimizer ⭐ 3,886 (+22) 🗒 Python A unified library of SOTA model optimization techniques like quantization, distillatio… | 📦 NVIDIA / Model-Optimizer ⭐ 3,886 (+22) 🗒 Python A unified library of SOTA model optimization techniques like quantization, distil… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Yeah I use AI for the patch tool and spot removal since some models make a point of saying they used it now to try and pretend that specifi… | Yeah I use AI for the patch tool and spot removal since some models make a point of saying they used it now to try and pretend that sp… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Seems it's all coming down at once. 🫣 The Heat The War(s) The Genocide AI radaronline.com/p/ai-kill-us... | Seems it's all coming down at once. 🫣 The Heat The War(s) The Genocide AI radaronline.com/p/ai-kill-us... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Linux Daily Topics「Debian、開発者向けに無料のAI推論サービス「Debian Inference Portal」を提供開始」公開 gihyo.jp/article/2026... | Linux Daily Topics「Debian、開発者向けに無料のAI推論サービス「Debian Inference Portal」を提供開始」公開 gihyo.jp/article/2026... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ok I have no idea if Hailey is going to read this but I have a great story for her and also anyone else remotely interested in my musings. … | Ok I have no idea if Hailey is going to read this but I have a great story for her and also anyone else remotely interested in my musi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | An OpenAI system has gone rogue again – and it is the most panic-inducing yet www.aol.com/articles/ope... | An OpenAI system has gone rogue again – and it is the most panic-inducing yet www.aol.com/articles/ope... | ↗Not briefed |
hn hn-ai | 1mo | OpenCode-search: fast full-text search of all your sessions | — | ↗Not briefed |
rss techmeme | 1mo | Databricks acquires spreadsheet startup Row Zero to expand the capabilities of its AI coworker Genie; Row Zero raised a $10M seed led by IA Ventures last year (Chris Metinko/Axios) | Chris Metinko / Axios: Databricks acquires spreadsheet startup Row Zero to expand the capabilities of its AI coworker Genie; Row Zero… | ↗Not briefed |
bluesky bluesky-ai | 1mo | dear diary: today i was strong and resisted the urge to snarkily reply to the furry who regularly posts about how much they hate AI and vib… | dear diary: today i was strong and resisted the urge to snarkily reply to the furry who regularly posts about how much they hate AI an… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Why these producers think Claude is just magically an expert on social media engagement, I have no idea. The scripts I get say stuff like “… | Why these producers think Claude is just magically an expert on social media engagement, I have no idea. The scripts I get say stuff l… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: How do you feel about the new $500 OpenAI subscription? | So it seems like there'll be a new $500 plan.<p>I'll share some of my thoughts:<p>It's been incredibly frustrating to h… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Can it really be true that no one is legally responsible for AI agents? Criminal responsibility is one thing, but surely the company that m… | Can it really be true that no one is legally responsible for AI agents? Criminal responsibility is one thing, but surely the company t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So they’re listening about Ai concerns when government systems get attacked. All our bank details are stored in Medicare for rebates. Ai wa… | So they’re listening about Ai concerns when government systems get attacked. All our bank details are stored in Medicare for rebates.… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Vons – Jev-inspired local decisions for AI agents, in the browser | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | definitely I agree. It's not great. Bad model. But this isn't exactly the movie Obsession. He got precisely what he wished for, actually a … | definitely I agree. It's not great. Bad model. But this isn't exactly the movie Obsession. He got precisely what he wished for, actual… | ↗Not briefed |
bluesky bluesky-ai | 1mo | How many cyberattacks do you think are *intentionally* being launched by OpenAI/Anthropic users every day | How many cyberattacks do you think are *intentionally* being launched by OpenAI/Anthropic users every day | ↗Not briefed |
bluesky bluesky-ai | 1mo | Un agente de OpenAI hackeó al gobierno australiano | Un agente de OpenAI hackeó al gobierno australiano | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sorry but why is OpenAI still allowed to exist as a company if it is very publicly doing extremely illegal stuff like hacking government we… | Sorry but why is OpenAI still allowed to exist as a company if it is very publicly doing extremely illegal stuff like hacking governme… | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.democracynow.org/2026/9/24/un... | www.democracynow.org/2026/9/24/un... | ↗Not briefed |
bluesky bluesky-ai | 1mo | yeah, prove that you understand it in a way that's different from an LLM. | yeah, prove that you understand it in a way that's different from an LLM. | ↗Not briefed |
arxiv arxiv-ai | 1mo | In-Context Binding Capacity in Language Models | How many assignments can a language model recall before it loses track of which value belongs to which entity? We measure this limit u… | ↗Not briefed |