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 | 29d | OpenAI says that the FBI arrested one of its employees in connection with a sophisticated listening device found hidden in one of the compa… | OpenAI says that the FBI arrested one of its employees in connection with a sophisticated listening device found hidden in one of the… | ↗Not briefed |
hn hn-ai | 29d | Light Touch: a classic iOS device emulator (iOS 1–7) | — | ↗Not briefed |
rss deepmind-blog | 29d | Google DeepMind and Singapore Government announce new AI tool to protect citizens against scams | Learn how Google DeepMind and GovTech Singapore uses lightweight on-device Gemma models to intercept conversational scams in private c… | ↗Not briefed |
arxiv arxiv-ai | 29d | W-JEPA: Wavelet Joint-Embedding Predictive Architecture for Wearable Activity Recognition | A major challenge in wearable device activity recognition is the heterogeneity across datasets in sensor placement, channel count, sam… | ↗Not briefed |
arxiv arxiv-ai | 29d | ForkGRPO: Efficient Exploration and Fine-Grained Credit Assignment for Diffusion Language Models | Diffusion large language models (dLLMs) have advanced rapidly, yet strengthening their reasoning through reinforcement learning (RL) r… | ↗Not briefed |
arxiv arxiv-ai | 29d | Vision-Language Models Integrate Conflicting Evidence Like Biased Bayesian Observers | Vision-language models must reconcile conflicting text and images. Prior work has characterized the resulting behavior in terms of mod… | ↗Not briefed |
bluesky bluesky-ai | 29d | If you are using the summaries at the top of a Google search, these are generated by AI aka LLM aka ChatGPT like things.🔧 ChatGPT | If you are using the summaries at the top of a Google search, these are generated by AI aka LLM aka ChatGPT like things. | ↗Not briefed |
arxiv arxiv-ai | 29d | Scaling LLM Inference via Parallelism: Throughput-Latency Trade-off | Large Language Model (LLM) serving systems must balance latency and throughput while efficiently utilizing distributed GPU resources.… | ↗Not briefed |
hn hn-ai | 29d | Nvidia Reportedly Ending the RTX 5090, Reserves GB202 for Workstation GPUs | — | ↗Not briefed |
arxiv arxiv-ai | 29d | ActMOF: Mapping Reticular Synthesis through Active Learning | Reticular chemistry allows metal-organic frameworks (MOFs) to be designed from their building units, yet their synthesis remains const… | ↗Not briefed |
bluesky bluesky-ai | 29d | An OpenAI engineer allegedly hid a mic wrapped in toilet paper in a room used for top-secret meetings. When he went looking for it, he was … | An OpenAI engineer allegedly hid a mic wrapped in toilet paper in a room used for top-secret meetings. When he went looking for it, he… | ↗Not briefed |
arxiv arxiv-ai | 29d | Train to Stitch: Topology-Preserving Regularization for Interoperable Representations | As large, powerful pre-trained models become more prevalent, there is growing interest in developing methods that allow different neur… | ↗Not briefed |
bluesky bluesky-ai | 29d | WSJ Reports Allegation That OpenAI Engineer Hid a Microphone Inside Toilet Paper in Conference Room | WSJ Reports Allegation That OpenAI Engineer Hid a Microphone Inside Toilet Paper in Conference Room | ↗Not briefed |
bluesky bluesky-ai | 29d | anthropic warns world could become “made of pudding” | anthropic warns world could become “made of pudding” | ↗Not briefed |
arxiv arxiv-ai | 29d | KoAGE: Do Large Language Models Know, Learn, and Coin Korean Neologisms? | Korean neologisms arise through diverse mechanisms, a demanding test of language models beyond lexical memorization. To enable systema… | ↗Not briefed |
arxiv arxiv-ai | 29d | SERO: Learning from Executable Oracle Continuations for Long-Horizon Agentic Reinforcement Learning | Long-horizon agentic RL is constrained by the futures visited by the current policy. When rollouts fail to reach useful behaviors, imp… | ↗Not briefed |
arxiv arxiv-ai | 29d | EpiGuard: An Autonomous Infectious-Disease Early Warning Agent Platform | Infectious disease outbreaks often spread rapidly before signs of widespread transmission become apparent, leaving a big challenge for… | ↗Not briefed |
bluesky bluesky-ai | 29d | you know how the ceo of openai and anthropic spoke at the UN about needing regulation around AI? well, the "godfather of AI" spoke about t… | you know how the ceo of openai and anthropic spoke at the UN about needing regulation around AI? well, the "godfather of AI" spoke ab… | ↗Not briefed |
arxiv arxiv-ai | 29d | Selective Next-Location Prediction with MetaModelSelect: A Lightweight Post-Hoc Confidence Metamodel | Next-location prediction can support proactive mobile-network decisions, but acting on unreliable predictions can waste resources. We… | ↗Not briefed |
bluesky bluesky-ai | 29d | like, anthropic is estimated to have, at the lowest end, 18M *consumer* subscribers, not enterprise | like, anthropic is estimated to have, at the lowest end, 18M *consumer* subscribers, not enterprise | ↗Not briefed |
arxiv arxiv-ai | 29d | Fully sequential Gaussian process regression for preference learning | We study continuous global optimization problems in which the objective function cannot be directly evaluated; rather, we observe only… | ↗Not briefed |
rss techmeme | 29d | Chinese manufacturers dominate humanoid robot production, capturing 97% of global shipments in H1 2026, thanks to government support and vast supply chains (Bloomberg) | Bloomberg: Chinese manufacturers dominate humanoid robot production, capturing 97% of global shipments in H1 2026, thanks to governmen… | ↗Not briefed |
bluesky bluesky-ai | 29d | 今、絵が崩れてると批判されてる漫画家の絵をLLMで添削するっていうとんでもない野蛮な文化が流行ってるみたいなんだ…人の絵を勝手に修整するのはゴーギャンとゴッホの間ですら許されなかったのに | 今、絵が崩れてると批判されてる漫画家の絵をLLMで添削するっていうとんでもない野蛮な文化が流行ってるみたいなんだ…人の絵を勝手に修整するのはゴーギャンとゴッホの間ですら許されなかったのに | ↗Not briefed |
arxiv arxiv-ai | 29d | ELARSpeech: A 30K-Hour Dataset Across 675 Underrepresented Languages for Long-Tail Multilingual Speech Technologies | Massively multilingual speech modeling has made remarkable progress in recent years, yet existing training resources and benchmarks st… | ↗Not briefed |
arxiv arxiv-ai | 29d | The Sum of All Summaries: PROPALIGN for Graded Proposition Salience Evaluation | Identification and ranking of salient propositions in documents is indirectly involved in many AI tasks, but few papers define proposi… | ↗Not briefed |
arxiv arxiv-ai | 29d | STRATA: Self-Maintaining Structured Experience Graphs for Self-Evolving Language Agents | Large language model (LLM) agents can self-evolve at test time by accumulating experience: retrieving relevant memories before acting… | ↗Not briefed |
arxiv arxiv-ai | 29d | RA$^2$: Retain-Anchored Attraction Preserves Forget-Adjacent Utility in LLM Unlearning | Large Language Model (LLM) unlearning is often evaluated primarily by target suppression and clean retain utility. These metrics can m… | ↗Not briefed |
rss gary-marcus | 29d | It remains “very difficult [to find] actual, materially measurable financial impacts” from spending on AI, per new study | Yet another study pointing in the same direction | ↗Not briefed |
arxiv arxiv-ai | 29d | ixi-sLM: On-Device sLM for AI Call Agent Service | Running an AI call agent on-device makes it private, network-independent, and free of per-query cloud cost, but shipping one as a real… | ↗Not briefed |
bluesky bluesky-ai | 29d | no, I have no idea what this means but I'm feeding it to _another_ LLM and asking it to translate and I'll share my notes | no, I have no idea what this means but I'm feeding it to _another_ LLM and asking it to translate and I'll share my notes | ↗Not briefed |
arxiv arxiv-ai | 29d | EvoWeave: Composing Reusable Agent Experience through Operation Level Graphs | Different tasks often require interactive agents to perform similar operations at intermediate steps. Retrieving an entire past workfl… | ↗Not briefed |
arxiv arxiv-ai | 29d | Learning What to Keep: Transfer-Aware Personal Skill Learning from User Feedback | User feedback enables LLMs to refine reusable personalized skills, but often mixes persistent preferences with request-specific correc… | ↗Not briefed |
bluesky bluesky-ai | 29d | i cannot fathom a fucking reason this is necessary, lmao. | i cannot fathom a fucking reason this is necessary, lmao. | ↗Not briefed |
bluesky bluesky-ai | 29d | see anthropic wouldnt need a custom git impl if they weren't slopping their entire codebase. | see anthropic wouldnt need a custom git impl if they weren't slopping their entire codebase. | ↗Not briefed |
arxiv arxiv-ai | 29d | Robust Explanatory Analysis for Structured Optimization Networks (REASON) in Learning | We seek a hierarchical resolution of residual error under a composite learning objective as a principled mechanism for identifying and… | ↗Not briefed |
bluesky bluesky-ai | 29d | OpenAI従業員、役員会議室に録音装置を仕掛けた疑いでFBIが逮捕 | OpenAI従業員、役員会議室に録音装置を仕掛けた疑いでFBIが逮捕 | ↗Not briefed |
bluesky bluesky-ai | 29d | Claudeさん、中国差別AIだった…「中国からのアクセスを検知しました、アカウントをBANしました」 https://ultchan.com/archives/302509 | Claudeさん、中国差別AIだった…「中国からのアクセスを検知しました、アカウントをBANしました」 https://ultchan.com/archives/302509 | ↗Not briefed |
bluesky bluesky-ai | 29d | 数年前から世界中でテクノロジーに明るい方々が危惧し警鐘を鳴らしてきた事が次々と的中する今日、安全性度外視のGenAIの設計不良によるサイバー犯罪や事故が毎日の様に報道されるようになってから、AnthropicやOpenAIなどGAI事業の関係者たちはGAI事業者の抱える課題(負… | 数年前から世界中でテクノロジーに明るい方々が危惧し警鐘を鳴らしてきた事が次々と的中する今日、安全性度外視のGenAIの設計不良によるサイバー犯罪や事故が毎日の様に報道されるようになってから、AnthropicやOpenAIなどGAI事業の関係者たちはGAI事業者の抱え… | ↗Not briefed |
arxiv arxiv-ai | 29d | DeLoop: Repeated Actions as Self-Supervision for Long-Horizon Agents | Language-model agents that run long tasks often redo work they have already done: they re-issue a tool call they made earlier and get… | ↗Not briefed |
bluesky bluesky-ai | 29d | We paid the bills. Your data has value. Anyone using twitter now, even lurking + occasional dm chats, is training Grok. DMs train Grok. You… | We paid the bills. Your data has value. Anyone using twitter now, even lurking + occasional dm chats, is training Grok. DMs train Grok… | ↗Not briefed |
hn hn-ai | 29d | Show HN: PDF to ePub Conversion Tool for Mac | I read a lot on e-ink readers. Lately I switched to a very small e-reader and PDFs are ~useless on it. So I vibe coded out a quick Swi… | ↗Not briefed |
arxiv arxiv-ai | 29d | One-Step Prediction Fidelity Is a Poor Proxy for Planning Performance in JEPA World Models🔧 Replicate | Latent world models are routinely selected, checkpointed and ablated on one-step prediction loss, on the assumption that a model which… | ↗Not briefed |
hn hn-ai | 29d | Apples to Apples: Getting Type Mappings Right in LLM-Based Code Translation | — | ↗Not briefed |
arxiv arxiv-ai | 29d | Evaluating Compressed Memory: Dependence, Utility, and Content Benefit | Does a language model answer better because its compressed memory contains useful document information? Replacing memory vectors with… | ↗Not briefed |
arxiv arxiv-ai | 29d | Policy-Verified Self-Training: Learning Agent Compliance from Verifiable Policy Rewards | Runtime policy verifiers can prevent agents from executing disallowed actions, but their decisions are typically used only for enforce… | ↗Not briefed |
hn hn-ai | 29d | KBEmacs, MIT's Programmer's Apprentice, Reanimated | — | ↗Not briefed |
hn hn-ai | 29d | Ask HN: Where has creativity in software gone? | I sometimes read Show HN, and there is literally no creativity at all: note-taker app after note-taker clone. Everyone is essentially… | ↗Not briefed |
arxiv arxiv-ai | 29d | The Impact of Detector Architecture, Model Scale, and Scene Density on Multi-Pedestrian Tracking | Pedestrian tracking in video is important for applications such as surveillance and scene analysis, but it becomes more challenging in… | ↗Not briefed |
arxiv arxiv-ai | 29d | Beyond Corrected Actions: Training LLM Agents on Textual Critiques | Large language models (LLMs) are increasingly used for multi-turn agentic tasks that require interactions with users and environments… | ↗Not briefed |
arxiv arxiv-ai | 29d | Seeing, Maintaining, and Learning: The Evolution of Multimodal Agent Memory | Multimodal models and agents have emerged to enable AI systems to learn from and interact with the real world more like humans by inte… | ↗Not briefed |