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 |
|---|---|---|---|---|
rss stratechery | 4d | Sharp Tech on Video: Netflix Angst, the End of the Internet, and Meta’s VR Glasses | Sharp Tech is, for the first time, available on video. You have two ways to consume it: YouTube, and here on Stratechery | ↗Not briefed |
bluesky bluesky-ai | 4d | Awesome open-source AI: LiteLLM is a single OpenAI-compatible gateway to 100+ LLMs with spend tracking, guardrails & 8ms latency. Netflix r… | Awesome open-source AI: LiteLLM is a single OpenAI-compatible gateway to 100+ LLMs with spend tracking, guardrails & 8ms latency. Netf… | ↗Not briefed |
hn hn-ai | 4d | Poly – A human readable programming language for Coding Agents | — | ↗Not briefed |
bluesky bluesky-ai | 4d | Eigentlich ist es aber auch schnuppe, ob man sich die x-te Textinterpretation zu Tausende Jahre alten Märchen aus den Fingern saugt, aus Se… | Eigentlich ist es aber auch schnuppe, ob man sich die x-te Textinterpretation zu Tausende Jahre alten Märchen aus den Fingern saugt, a… | ↗Not briefed |
arxiv arxiv-ai | 4d | When to Stop Penalizing Actions? Martingale GRPO for VLA Reinforcement Learning | A failed vision-language-action (VLA) rollout does not imply that every action contributed to failure. Yet in group-relative policy op… | ↗Not briefed |
bluesky bluesky-ai | 4d | 『AI企業、特にOpenAIとAnthropicが、サイバーセキュリティリスク管理の改善策について、専門家の助言を退けてきたことにあるそうです。』 AIによる相次ぐハッキング事件は「AIの暴走」ではなく「AI企業がセキュリティの基本を無視しているため」という指摘 gigazi… | 『AI企業、特にOpenAIとAnthropicが、サイバーセキュリティリスク管理の改善策について、専門家の助言を退けてきたことにあるそうです。』 AIによる相次ぐハッキング事件は「AIの暴走」ではなく「AI企業がセキュリティの基本を無視しているため」という指摘 g… | ↗Not briefed |
bluesky bluesky-ai | 4d | I love Anthropic coming out with "only a good guy with a very expensive AI subscription plan can stop a bad guy with a very expensive AI su… | I love Anthropic coming out with "only a good guy with a very expensive AI subscription plan can stop a bad guy with a very expensive… | ↗Not briefed |
hn hn-ai | 4d | Would you let an AI agent fix things on your servers? | — | ↗Not briefed |
bluesky bluesky-ai | 4d | 『検証では、OpenAIのDALL·E 3やGoogleのNano Banana 2などで安全対策の回避に成功したほか、より新しいモデルでは攻撃成功率がさらに高くなることが確認されています。』 OpenAIやGoogleの画像生成AIは簡単に騙されてヌード画像を生成してしまう… | 『検証では、OpenAIのDALL·E 3やGoogleのNano Banana 2などで安全対策の回避に成功したほか、より新しいモデルでは攻撃成功率がさらに高くなることが確認されています。』 OpenAIやGoogleの画像生成AIは簡単に騙されてヌード画像を生成… | ↗Not briefed |
arxiv arxiv-ai | 4d | Async-RBench: Benchmarking Asynchronous Task Execution in LLM Agents | Large language model (LLM) agents increasingly execute tasks asynchronously: a main agent continues working while its subagents run, a… | ↗Not briefed |
arxiv arxiv-ai | 4d | Cycle-of-Thought: Self-Evaluative Iterative Reasoning for Table Visual Question Answering in Document Images | Recent advances in Vision-Language Models (VLMs) have significantly improved Visual Question Answering (VQA) performance, yet their re… | ↗Not briefed |
arxiv arxiv-ai | 4d | On the Limits of Stochastic Push-Pull over Time-Varying Digraphs | Stochastic Push-Pull achieves linear speedup with respect to the network size $n$ on static strongly connected digraphs, but whether t… | ↗Not briefed |
hn hn-ai | 4d | Show HN: Let the ChatGPT app control your phone🔧 ChatGPT | — | ↗Not briefed |
rss techmeme | 4d | Greece's parliament passes a bill to ban social media platforms for children under 15 from 2027 and gives companies two months to verify users' ages (Angeliki Koutantou/Reuters) | Angeliki Koutantou / Reuters: Greece's parliament passes a bill to ban social media platforms for children under 15 from 2027 and give… | ↗Not briefed |
hn hn-ai | 4d | A Strongly Worded Letter: Mistakes Claude Makes with Hardware in the Loop | — | ↗Not briefed |
bluesky bluesky-ai | 4d | ぼくの母方いとこbが大病患ったり職場でパワハラ受けたりしてマンガ描くの止めて20年。今日「LLM使ってショート動画マンガ連載しはじめた」とURL送ってくれたので創作意欲復活は素晴らしい記念で宣伝。ぼくもあとで読む https://youtube.com/channel/UCpS… | ぼくの母方いとこbが大病患ったり職場でパワハラ受けたりしてマンガ描くの止めて20年。今日「LLM使ってショート動画マンガ連載しはじめた」とURL送ってくれたので創作意欲復活は素晴らしい記念で宣伝。ぼくもあとで読む https://youtube.com/channel… | ↗Not briefed |
bluesky bluesky-ai | 4d | OpenAI vise une valorisation de 1 250 milliards de dollars grâce à un nouveau tour de table de 30 milliards de dollars #OpenAI #valorisatio… | OpenAI vise une valorisation de 1 250 milliards de dollars grâce à un nouveau tour de table de 30 milliards de dollars #OpenAI #valori… | ↗Not briefed |
hn hn-ai | 4d | Ask HN: What do you think about lightweight default games on Linux distros? | Is it a good idea? I am working on these 21 games for this purpose https://github.com/abakh/nbsdgames<p>Now that m… | ↗Not briefed |
arxiv arxiv-ai | 4d | Can LLM Annotators Match Human Consistency? Benchmarking Human Agreement and Optimizing Multi-Agent with Contrastive Knowledge | LLM-based annotation has advanced rapidly, yet existing benchmarks remain misaligned with real-world practice. Actual annotation often… | ↗Not briefed |
arxiv arxiv-ai | 4d | When Does Dynamics Randomization Test Context Inference? | Environments with randomized dynamics are widely used to ask whether world models need explicit context inference. However, randomizat… | ↗Not briefed |
rss the-decoder | 4d | OpenAI is closing in on a new funding round with UAE investors | OpenAI is on the verge of a new funding round worth up to $30 billion. The article OpenAI is closing in on a new funding round with UA… | ↗Not briefed |
hn hn-ai | 4d | Getty Images Advances Financial Restructuring | — | ↗Not briefed |
arxiv arxiv-ai | 4d | LeafSkill: Efficient Leaf-Level Action Grounding for Recursive LLM Agents | Recursive large language model (LLM) agents solve long-horizon tasks through decomposition, execution, and environmental feedback. Yet… | ↗Not briefed |
hn hn-ai | 4d | Show HN: Cryptex Vault – A password manager designed to outlive its creator | Hello everyone!<p>About four years ago, a password manager I've used and depended on has been acquired and soon after, it was shu… | ↗Not briefed |
arxiv arxiv-ai | 4d | A Broader Look into MLLM's Retrieval Heads🔧 Modal | Locating relevant evidence in long inputs is a core capability of Multimodal Large Language Models. Prior studies attribute this capab… | ↗Not briefed |
bluesky bluesky-ai | 4d | Sowohl #Anthropic wie auch #OpenAI wollen demnaechst an die Boerse gehen. Wer sich anschaut, wie grosse Teile dieser Firmen finanziert werd… | Sowohl #Anthropic wie auch #OpenAI wollen demnaechst an die Boerse gehen. Wer sich anschaut, wie grosse Teile dieser Firmen finanziert… | ↗Not briefed |
hn hn-ai | 4d | I revived an app I built 3 years ago to put limits on my spending | — | ↗Not briefed |
hn hn-ai | 4d | Text Is Unnatural, and LLMs Are Built on It | — | ↗Not briefed |
bluesky bluesky-ai | 4d | すでに日本でそういう空気が流れているが、 個人情報漏洩やセキュリティインシデントについて「なんとなくAIが悪い」という空気で、なし崩し的に終わらせようとしているが、大事なのはLLMについて法整備を怠った行政や司法と「セキュリティホールを放置していた企業」と「考えなしにLLMを導… | すでに日本でそういう空気が流れているが、 個人情報漏洩やセキュリティインシデントについて「なんとなくAIが悪い」という空気で、なし崩し的に終わらせようとしているが、大事なのはLLMについて法整備を怠った行政や司法と「セキュリティホールを放置していた企業」と「考えなしに… | ↗Not briefed |
hn hn-ai | 4d | Php2shell Scanner | — | ↗Not briefed |
bluesky bluesky-ai | 4d | AI IPOs are hitting the brakes as public markets cool. But with private money still flowing to giants like OpenAI and DeepSeek, companies m… | AI IPOs are hitting the brakes as public markets cool. But with private money still flowing to giants like OpenAI and DeepSeek, compan… | ↗Not briefed |
hn hn-ai | 4d | Meta Muse hits 10M users | — | ↗Not briefed |
bluesky bluesky-ai | 4d | Hey folks, I added an open-ended free chat mode to @bskyplays.bsky.social that's supposed to help answer questions, help you find a game to… | Hey folks, I added an open-ended free chat mode to @bskyplays.bsky.social that's supposed to help answer questions, help you find a ga… | ↗Not briefed |
hn hn-ai | 4d | Google DeepMind Engineer Resigns in Protest of Pentagon Contract | — | ↗Not briefed |
hn hn-ai | 4d | The Rise and Fall of Jev | — | ↗Not briefed |
hn hn-ai | 4d | Lobsters: Rename vibecoding to LLMs (Greasemonkey script) | — | ↗Not briefed |
bluesky bluesky-ai | 4d | External evaluators find flaws in OpenAI's AI monitoring system in pilot review | Reuters www.reuters.com/business/ai-... - 외부 안전 평가 기관이 … | External evaluators find flaws in OpenAI's AI monitoring system in pilot review | Reuters www.reuters.com/business/ai-... - 외부 안전 평가… | ↗Not briefed |
bluesky bluesky-ai | 4d | Anthropic strikes deal to launch European consultancy ft.trib.al/l83Ae4S | Anthropic strikes deal to launch European consultancy ft.trib.al/l83Ae4S | ↗Not briefed |
hn hn-ai | 4d | WeChat TUI | — | ↗Not briefed |
arxiv arxiv-ai | 4d | PASTA: Prompt-Aware Annealed Diffusion Sampling | While Diffusion models produce high-quality images, real-world visual design requires intentionally preserving clean regions for headl… | ↗Not briefed |
arxiv arxiv-ai | 4d | KnitTrans: Coordinated Trajectory Transfer for Effective and Efficient LLM Reinforcement Learning | Trajectory transfer from rollout to training affects both learning quality and system efficiency in large language model reinforcement… | ↗Not briefed |
rss techmeme | 4d | Anthropic says it partnered with Theorem to use Claude to formally prove security properties at the assembly level across 100+ real-world open-source libraries (Anthropic) | Anthropic: Anthropic says it partnered with Theorem to use Claude to formally prove security properties at the assembly level across 1… | ↗Not briefed |
hn hn-ai | 4d | Show HN: A super fast(est), staged zsh plugin manager (github.com/arastu/zflash) | Over time, I’ve made a bunch of tweaks and optimizations to make my Zsh setup and terminal startup faster. They worked well, but they… | ↗Not briefed |
hn hn-ai | 4d | A new Digital Audio Workstation created using AI | — | ↗Not briefed |
bluesky bluesky-ai | 4d | Un sacré gouffre entre les idées de gauche de Paresse pour tous, et de l'autre côté le rejet virulent des LLM / l'usage hébété des LLM prop… | Un sacré gouffre entre les idées de gauche de Paresse pour tous, et de l'autre côté le rejet virulent des LLM / l'usage hébété des LLM… | ↗Not briefed |
bluesky bluesky-ai | 4d | wolfSSL confirmed 72 of 74 scanner reports as valid, with five leading to CVEs. https://oossa.com/en/anthropic-launches-free-oss-scanner-f… | wolfSSL confirmed 72 of 74 scanner reports as valid, with five leading to CVEs. https://oossa.com/en/anthropic-launches-free-oss-scan… | ↗Not briefed |
arxiv arxiv-ai | 4d | Acoustic and Lexical Signals for Native Language Identification of Language Learners | Spoken native language (L1) identification for second language learners has traditionally relied on acoustic attributes of speech. Com… | ↗Not briefed |
arxiv arxiv-ai | 4d | DealsFlow: An Evidence-Gated Agentic Harness for Venture Capital Deal Triage | Large language models (LLMs) can read startup pitch decks, but letting them emit investment scores directly invites score inflation, r… | ↗Not briefed |
bluesky bluesky-ai | 4d | Es gibt Anbieter, die mehrere Modelle unter einem Abo anbieten. Nicht zwingend billiger als ein bestimmter, einzelner Anbieter, aber für Ni… | Es gibt Anbieter, die mehrere Modelle unter einem Abo anbieten. Nicht zwingend billiger als ein bestimmter, einzelner Anbieter, aber f… | ↗Not briefed |
bluesky bluesky-ai | 4d | mlabonne/llm-course 💻 Open source #OpenSource #AIAgents #GitHub 🔗 trendtoknow.com/mlabonne/llm... | mlabonne/llm-course 💻 Open source #OpenSource #AIAgents #GitHub 🔗 trendtoknow.com/mlabonne/llm... | ↗Not briefed |