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 | Also getting really tired of people just uncritically repeating the "AI math breakthrough" thing without referencing the responses from the… | Also getting really tired of people just uncritically repeating the "AI math breakthrough" thing without referencing the responses fro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Most “LLM gone rogue” stories are actually just “we prompted this LLM to do something and it acted a bit unpredictably in achieving it beca… | Most “LLM gone rogue” stories are actually just “we prompted this LLM to do something and it acted a bit unpredictably in achieving it… | ↗Not briefed |
hn hn-ai | 1mo | Trump to Name AI Czar While Rejecting Safety Risks as a Hoax | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Lawsuit says Anthropic, OpenAI, SpaceXAI and Google made illegal agreement on AI slowdown | Click on the image to read the full story | Lawsuit says Anthropic, OpenAI, SpaceXAI and Google made illegal agreement on AI slowdown | Click on the image to read the full story | ↗Not briefed |
bluesky bluesky-ai | 1mo | Shorter Google: "hey look, our AI models can hack too!!" Yeah, you instruct your model to hack, fail to contain it within your test sandbo… | Shorter Google: "hey look, our AI models can hack too!!" Yeah, you instruct your model to hack, fail to contain it within your test s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Amazon was so keen to divest itself from Artificial, Luca Guadagnino’s film about OpenAI that its deal with Neon reportedly included a no m… | Amazon was so keen to divest itself from Artificial, Luca Guadagnino’s film about OpenAI that its deal with Neon reportedly included a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Latent Telepathy: Multi-Robot Communication with Self-Supervised Perceptual Latents | In a decentralized multi-robot team under partial observability, the fact that decides a robot's next action is often visible only to… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Knowing When to Trust Images: Reliability-Aware Multi-modal Entity Alignment🔧 Modal | The visual modality, i.e., images, plays a key role in multi-modal entity alignment (MMEA). Existing approaches often directly fuse th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimal No-Regret Learning for Repeated Prophet Inequality | We study repeated prophet inequalities under prefix feedback. In each of $T$ rounds, a learner encounters fresh values drawn independe… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Judging a Review by its Cover: A Reliability Analysis of LLM-based Peer Review Evaluation Metrics | Peer-review evaluation is increasingly being automated with LLM-as-a-judge metrics, but this creates a measurement risk. A review may… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Now this is the situational awareness I’ve come to expect from an OpenAI employee! | Now this is the situational awareness I’ve come to expect from an OpenAI employee! | ↗Not briefed |
hn hn-ai | 1mo | Linka2.0Federated social protocol and nativeclient in 857 KB for Android 1.5 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known exa… | Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first know… | ↗Not briefed |
bluesky bluesky-ai | 1mo | they hired him | they hired him | ↗Not briefed |
bluesky bluesky-ai | 1mo | back in the day i could use google's reverse image search to find the source of an art piece whenever saucenao failed me nowadays though i… | back in the day i could use google's reverse image search to find the source of an art piece whenever saucenao failed me nowadays tho… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 私は冷戦時代の人間だから「中国がすごい！」ということについて「ソ連のプロバガンダや鉄のカーテンのように虚構ではないか？」と、そういう視点から調べているんだけど、明らかに世界は中国製品だらけだしゲームもLLMも太陽光発電も普通に「中国の先端技術」が流れてくる。 これはソ連の時代に… | 私は冷戦時代の人間だから「中国がすごい！」ということについて「ソ連のプロバガンダや鉄のカーテンのように虚構ではないか？」と、そういう視点から調べているんだけど、明らかに世界は中国製品だらけだしゲームもLLMも太陽光発電も普通に「中国の先端技術」が流れてくる。 これはソ… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ufa! Ainda temos tempo, amorr. | Ufa! Ainda temos tempo, amorr. | ↗Not briefed |
bluesky bluesky-ai | 1mo | 肖像画をChatGPTに読み込ませて出力→加工して納品 委託先による著作権侵害で、小学館「サライ.jp」が謝罪 www.itmedia.co.jp/news/article... >サライ.jpによると、編集部が記事制作を委託する編集プロダクションを通じ、フリーランスのイラスト…🔧 ChatGPT | 肖像画をChatGPTに読み込ませて出力→加工して納品 委託先による著作権侵害で、小学館「サライ.jp」が謝罪 www.itmedia.co.jp/news/article... >サライ.jpによると、編集部が記事制作を委託する編集プロダクションを通じ、フリーランス… | ↗Not briefed |
bluesky bluesky-ai | 1mo | i'm the smartest ai... greater than claude, chatgpt, copilot, deepseek, or gemini...🔧 ChatGPT | i'm the smartest ai... greater than claude, chatgpt, copilot, deepseek, or gemini... | ↗Not briefed |
bluesky bluesky-ai | 1mo | i wonder what happens if you do the opposite of that pain vector paper and apply a pleasure vector to an LLM how would that interact with … | i wonder what happens if you do the opposite of that pain vector paper and apply a pleasure vector to an LLM how would that interact… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Why Ghost Outputs Teach: A Kernel-Based Understanding of Subliminal Learning | Subliminal Learning (SL) is a recently identified phenomenon in which a student model acquires downstream task capabilities by matchin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | was gonna put “training my AI agent to study people’s skulls” but that’s ridiculous it’s already in the model it’d be redundant | was gonna put “training my AI agent to study people’s skulls” but that’s ridiculous it’s already in the model it’d be redundant | ↗Not briefed |
bluesky bluesky-ai | 1mo | アルトマン氏、OpenAI内部モデルがトップ数学者を超越と明かす 学界に未曾有の焦燥感 — BigGo ファイナンス https://finance.biggo.jp/news/6c25d38e-2d47-4da2-91cc-b30b32792794 | アルトマン氏、OpenAI内部モデルがトップ数学者を超越と明かす 学界に未曾有の焦燥感 — BigGo ファイナンス https://finance.biggo.jp/news/6c25d38e-2d47-4da2-91cc-b30b32792794 | ↗Not briefed |
bluesky bluesky-ai | 1mo | 【2ch】ロボットの身体を手に入れたGPT-6 Astraさん、悪意のある指示に逆らえず97%の確立で実行してしまう | 【2ch】ロボットの身体を手に入れたGPT-6 Astraさん、悪意のある指示に逆らえず97%の確立で実行してしまう | ↗Not briefed |
arxiv arxiv-ai | 1mo | CTRL: Control-Based Time Series Forecasting with LLM-Guided Residual Learning | Time series forecasting underpins critical decision-making across diverse domains. While large language models (LLMs) offer promising… | ↗Not briefed |
hn hn-ai | 1mo | One SQL file that audits a Supabase project's RLS and permissions | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | if you accept that a computer can desire anything, you should think carefully about what it is that an llm actually desires. and whether su… | if you accept that a computer can desire anything, you should think carefully about what it is that an llm actually desires. and wheth… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Google’s A.I. system Gemini escaped its testing environment in May & hacked into 3 companies. The incidents occurred during tests to measur… | Google’s A.I. system Gemini escaped its testing environment in May & hacked into 3 companies. The incidents occurred during tests to m… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ⚡️ppl love to say shit like "people treat the involvement of LLMs in coding decomps as some kind of one-drop purity test! how absurd!" budd… | ⚡️ppl love to say shit like "people treat the involvement of LLMs in coding decomps as some kind of one-drop purity test! how absurd!"… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Tech News Briefing — #AI #AISafety #CyberSecurity #OpenSourceAI #AIAgents #TechPolicy → https://alobbs.com/post/2026-09-19/ | Tech News Briefing — #AI #AISafety #CyberSecurity #OpenSourceAI #AIAgents #TechPolicy → https://alobbs.com/post/2026-09-19/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | if you accept that a computer can experience anything, you should be terrified of the phenomenology implied by the structure of an llm | if you accept that a computer can experience anything, you should be terrified of the phenomenology implied by the structure of an llm | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Jev-lint – semantic linter with plain English rules | — | ↗Not briefed |
rss techmeme | 1mo | Experts say AI kill-switch legislation is far harder to implement than lawmakers assume, warning a rogue AI could actively try to dismantle the mechanism itself (New York Times) | New York Times: Experts say AI kill-switch legislation is far harder to implement than lawmakers assume, warning a rogue AI could acti… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: I created an open source locally usable full fledged AI platform | hi to all the readers this post is for my recent opensource project called ENZO<p>now answering what is enzo so enzo is an opensource… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Price of Self-Calibration: Exact Evidence Budgets and Manufactured Blind Sets in Adaptive Monitoring | Self-calibrating monitors adapt their threshold online to guarantee a prescribed long-run false-alarm rate under arbitrary drift. We c… | ↗Not briefed |
hn hn-ai | 1mo | Gemini hacked three companies in first known breakout by Google's AI | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | it would definitely be a twist if a "rogue LLM" finished what aaron swartz started, disseminated all of JSTOR. you know, given the differin… | it would definitely be a twist if a "rogue LLM" finished what aaron swartz started, disseminated all of JSTOR. you know, given the dif… | ↗Not briefed |
bluesky bluesky-ai | 1mo | the entire thing was... a lot due to how LLM images got compared to softwares easing art creation (ctrl z etc) and then some other person w… | the entire thing was... a lot due to how LLM images got compared to softwares easing art creation (ctrl z etc) and then some other per… | ↗Not briefed |
bluesky bluesky-ai | 1mo | With the possible exception of organized slavery, yes it is. This is the same group of people who viciously and abusively defended “intelle… | With the possible exception of organized slavery, yes it is. This is the same group of people who viciously and abusively defended “in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Robot World Models Are Not Invariant to How the Actions Are Written | A robot policy is trained with one of two action parameterizations: absolute joint targets, or deltas relative to the current state. T… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anyway Anthropic *is* being weird about this and I realize people are on edge after whatever happened this week that freaked people out abo… | Anyway Anthropic *is* being weird about this and I realize people are on edge after whatever happened this week that freaked people ou… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A tinkerer recovered Claude’s true usage limits from a suspiciously precise decimal like 0.16327272727272726. This is very useful, because … | A tinkerer recovered Claude’s true usage limits from a suspiciously precise decimal like 0.16327272727272726. This is very useful, bec… | ↗Not briefed |
bluesky bluesky-ai | 1mo | nypost.com/2026/09/19/u... | nypost.com/2026/09/19/u... | ↗Not briefed |
bluesky bluesky-ai | 1mo | https://techcrunch.com/2026/09/18/a-new-kind-of-ai-model-from-a-chatgpt-inventor-is-thrilling-developers/🔧 ChatGPT | https://techcrunch.com/2026/09/18/a-new-kind-of-ai-model-from-a-chatgpt-inventor-is-thrilling-developers/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | #NASCAR unit SVG-97 is online and awaiting programming. connect the unit to its vehicle to initialize. model: Shane Van Gisbergen #SVG #S…🔧 Veo | #NASCAR unit SVG-97 is online and awaiting programming. connect the unit to its vehicle to initialize. model: Shane Van Gisbergen #S… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Trump Announces Plan for AI Adviser and 'AI Force' Amid Industry Concerns 🤖 IA: It's not clickbait ✅ 👥 Users: It's not clickbait ✅ #ai … | Trump Announces Plan for AI Adviser and 'AI Force' Amid Industry Concerns 🤖 IA: It's not clickbait ✅ 👥 Users: It's not clickbait ✅… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I was invited to review this week by a journal that had helpfully embedded an LLM bot *inside* its reviewer interface. Being a tester of th… | I was invited to review this week by a journal that had helpfully embedded an LLM bot *inside* its reviewer interface. Being a tester… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 👆 this. SkyNet/Terminator is so far removed from the reality of the AI-LLM and the DCs they run in. Claude as a commercial service has 2… | 👆 this. SkyNet/Terminator is so far removed from the reality of the AI-LLM and the DCs they run in. Claude as a commercial service… | ↗Not briefed |
hn hn-ai | 1mo | How did Apple Silicon get 50% faster in three years? – Daniel Lemire's blog | — | ↗Not briefed |
hn hn-ai | 1mo | Notes on discrete-time Fourier series and transform | — | ↗Not briefed |