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 | 2mo | Steering Generative Robot Policies with Lexicographic Preferences | Pretrained generative robot policies can produce effective behaviors across diverse environments, but deployment can lead to requireme… | ↗Not briefed |
bluesky bluesky-ai | 2mo | www.repubblica.it/esteri/2026/... Mentre gli imprenditori del settore chiedono almeno una pausa di riflessione l’unico ad invocare di proce… | www.repubblica.it/esteri/2026/... Mentre gli imprenditori del settore chiedono almeno una pausa di riflessione l’unico ad invocare di… | ↗Not briefed |
hn hn-ai | 2mo | OpenAI Astra checks webcam for user | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | Overflip: Repetition-Induced Label Flips in Guardrail Models | Guardrail models are classifiers deployed to screen malicious prompts and responses in LLM-based services. To meet latency constraints… | ↗Not briefed |
rss bing-news-ai | 2mo | Latest updates as AI experts warn about imminent catastrophe🔧 ChatGPT | AI warnings - latest: Experts warn about imminent catastrophe as artificial intelligence companies agree to slow development - Leaders… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Student newspaper op ed: "Penn can’t hold true to its values of being “open-minded and curious, inventive and practical, exhibiting brilli… | Student newspaper op ed: "Penn can’t hold true to its values of being “open-minded and curious, inventive and practical, exhibiting b… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Global AI stocks slip on concerns over risks ft.trib.al/zLKUlIJ | Global AI stocks slip on concerns over risks ft.trib.al/zLKUlIJ | ↗Not briefed |
bluesky bluesky-ai | 2mo | I'm still pretty convinced it's an LLM at this point, it only has the one mode | I'm still pretty convinced it's an LLM at this point, it only has the one mode | ↗Not briefed |
arxiv arxiv-ai | 2mo | Semantic-TVM: Structure-Preserving Trustworthy Virtual Memory for Memory-Augmented and Tool-Using Agents | Memory-augmented and tool-using agents expose exact private values when remote LLMs process retrieved memory, tool actions, and interm… | ↗Not briefed |
arxiv arxiv-ai | 2mo | CoMem: Collective-Individual Memory Synergy for Evolutionary Multi-Agent Systems | Designing effective memory mechanisms is crucial for advancing LLM-driven Multi-Agent Systems (MAS), helping agents learn together and… | ↗Not briefed |
arxiv arxiv-ai | 2mo | ABSOL: Aggregated Bayesian Subsampling Orchestrated with LLMs🔧 GPT-5 | Large language models are increasingly used as natural-language interfaces to structured data, yet they remain unreliable when answers… | ↗Not briefed |
bluesky bluesky-ai | 2mo | 🚨 US-China AI Race vs. Global Safety Warnings: Where Does the US Stand? 🚨 Share your thoughts below! 👇 RMN News: rmnnews.com/2026/09/14/… | 🚨 US-China AI Race vs. Global Safety Warnings: Where Does the US Stand? 🚨 Share your thoughts below! 👇 RMN News: rmnnews.com/2026/0… | ↗Not briefed |
hn hn-ai | 2mo | The "omarchy" stuff is getting weird and weirder | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | IMPACT-VLA: Interaction-aware Multimodal Propagation Attribution via Counterfactual Trajectories for Vision-Language-Action Policies | Vision-Language-Action (VLA) policies perform robot manipulation tasks using multimodal inputs such as visual observations, propriocep… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Absolute Leseempfelung 👇 KI-Apokalypse - Wie gefaehrlich wird uns kuenstliche Intelligenz? @palle.eurosky.social 🙏 Klasse 👍 www.meta… | Absolute Leseempfelung 👇 KI-Apokalypse - Wie gefaehrlich wird uns kuenstliche Intelligenz? @palle.eurosky.social 🙏 Klasse 👍 www… | ↗Not briefed |
hn hn-ai | 2mo | Sam Altman: Frontier AI labs need shared safety rules and responsible pacing | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | Buddy, an LLM is liable to give me a bloodthirsty demonic Muramasa & get me executed by the dang shogun! | Buddy, an LLM is liable to give me a bloodthirsty demonic Muramasa & get me executed by the dang shogun! | ↗Not briefed |
bluesky bluesky-ai | 2mo | OpenAI finalmente da crédito por la solución propuesta a uno de los siete Problemas del Milenio. Incorporó el trabajo de Diego Córdoba y Lu… | OpenAI finalmente da crédito por la solución propuesta a uno de los siete Problemas del Milenio. Incorporó el trabajo de Diego Córdoba… | ↗Not briefed |
arxiv arxiv-ai | 2mo | HGTO: A Unified Graph-Based Physics-Informed Formulation for Structural Topology Optimization | Density-based topology optimization is typically structured as a nested sequence of material updates, structural analyses, and sensiti… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Anthropic walks tightrope to Nasdaq, pushing for a slowdown while pursuing $2 trillion valuation | Anthropic walks tightrope to Nasdaq, pushing for a slowdown while pursuing $2 trillion valuation | ↗Not briefed |
hn hn-ai | 2mo | EPA Expected to Erase Limits on Climate Pollution from Power Plants | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | Shallow Beliefs: Synthetic document finetuning does not inoculate against emergent misalignment from reward hacking | Recent work shows that models that learn to reward hack on RL environments can become broadly misaligned, and that reframing reward ha… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Beyond Depth and Width: The Information-Slack Dilemma in Streaming Test-Time Compute | The same task and compute budget can require different reasoning policies when evidence arrives in a different order. Early computatio… | ↗Not briefed |
bluesky bluesky-ai | 2mo | LLMと開発や設計など建設的な話をしていると、あまりにも建設的なので、人間の価値ってなんだろうなって考えるよね。 悲観的に見ているんじゃなくて、役割分担があるなと。 例えば人間には身体があって、人生があり、眠くなったり疲れたりする。大小さまざまな困り事があって、解決したいと… | LLMと開発や設計など建設的な話をしていると、あまりにも建設的なので、人間の価値ってなんだろうなって考えるよね。 悲観的に見ているんじゃなくて、役割分担があるなと。 例えば人間には身体があって、人生があり、眠くなったり疲れたりする。大小さまざまな困り事があって、解… | ↗Not briefed |
bluesky bluesky-ai | 2mo | GPT-6 AstraとClaude Fableの性能を引き出すハーネス「Fusion」が登場、Claude CodeやCodexで使う場合と同等の性能を示しつつコストを最大39%削減 https://gigazine.net/news/20260914-cognition-f… | GPT-6 AstraとClaude Fableの性能を引き出すハーネス「Fusion」が登場、Claude CodeやCodexで使う場合と同等の性能を示しつつコストを最大39%削減 https://gigazine.net/news/20260914-cognit… | ↗Not briefed |
arxiv arxiv-ai | 2mo | MTAC-IFBench: Benchmarking Instruction-Following in Multi-Turn Agentic Coding | Recently, the rapid development of large language models (LLMs) has reshaped software engineering by enabling autonomous code agents t… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Based on interviews across 24 countries in 7 regions of the world for a report from the Berkeley Human Rights Center, Betsy Popken and Vyom… | Based on interviews across 24 countries in 7 regions of the world for a report from the Berkeley Human Rights Center, Betsy Popken and… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Typhoon ASR Streaming: Steerable Low-Latency Thai Speech Recognition with Real-Time Shallow Fusion🔧 Whisper | Open Thai automatic speech recognition (ASR) is dominated by offline, Whisper-based models that read the whole utterance before transc… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Sakana AIが複数モデルの使い分けでGPT-6 Astra超えを達成する「Fugu Ultra v2」をリリース&コスパに優れた「Fugu Max」も登場 https://gigazine.net/news/20260914-fugu-ultra-v2/ | Sakana AIが複数モデルの使い分けでGPT-6 Astra超えを達成する「Fugu Ultra v2」をリリース&コスパに優れた「Fugu Max」も登場 https://gigazine.net/news/20260914-fugu-ultra-v2/ | ↗Not briefed |
arxiv arxiv-ai | 2mo | Learned Bow Control on a Measured Bowed-String Model: a Revised Minimum-Bow-Force Law, a Recurrent Controller, and the Domain of a Supervision Ceiling | A finite-difference bowed-string model with implicitly resolved Stribeck friction is presented, with a regime diagnostic, the Schellen… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Biomedical Reference Generation Remains Unreliable across 26 Large Language Models🔧 GPT-5 | Background. Large language models are increasingly used to help write biomedical text but may fabricate references to nonexistent work… | ↗Not briefed |
arxiv arxiv-ai | 2mo | ActGuard: Pre-execution Action Auditing against Indirect Prompt Injection in LLM Agents | Large language model (LLM) agents interact with external environments through tool invocation, but tool outputs can also expose them t… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Converting Sequenced Fuzzy Cognitive Maps to Causal Virtual Worlds with Large Video Generators🔧 Veo | We show how users can create and manipulate causal virtual worlds with large-language-model (LLM) and large-video-model agents. The ap… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Es drängt sich der Eindruck auf, dass Trump und seine Bande systematisch an allen Fronten daran arbeiten, der Erde den größtmöglichen Schad… | Es drängt sich der Eindruck auf, dass Trump und seine Bande systematisch an allen Fronten daran arbeiten, der Erde den größtmöglichen… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Anthropic Strikes $13.7 Billion Compute Deal With Trump-Linked Rum Group -- The Information Rum Group used to be known as Rumble. Peter Ti… | Anthropic Strikes $13.7 Billion Compute Deal With Trump-Linked Rum Group -- The Information Rum Group used to be known as Rumble. Pet… | ↗Not briefed |
arxiv arxiv-ai | 2mo | LiftGCN: Efficient Energy-Preserving Graph Learning via Joukowski Spectral Lifting for Finite Element Stress Prediction | Finite element stress fields often exhibit strong local non-smoothness, where stress concentrations near holes, notches, and loading r… | ↗Not briefed |
arxiv arxiv-ai | 2mo | MemRiskBench: Trace-Aware Risk-Preserving Evaluation for Long-Horizon LLM Agents | Long-horizon LLM agents accumulate memory across sessions, creating sparse but high-impact risks: stale facts, conflicting updates, cr… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Calls for regulation of AI coming from big tech companies should be treated with deep suspicion, especially when linked to a US program to … | Calls for regulation of AI coming from big tech companies should be treated with deep suspicion, especially when linked to a US progra… | ↗Not briefed |
bluesky bluesky-ai | 2mo | first off that solve was heavily based on the work of a lot of other mathematicians (who i think should share credit) - the llm that put it… | first off that solve was heavily based on the work of a lot of other mathematicians (who i think should share credit) - the llm that p… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Learning to Solve Stochastic Controls with Unknown Drifts and Running Rewards: Theory, Algorithms and Convergence | We study continuous-time and possibly high-dimensional stochastic control problems where drift coefficients and running reward functio… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Towards a knowledge-enhanced single-cell foundation model | Single-cell foundation models (scFMs) increasingly rely on large-scale transcriptomic pretraining, yet expanding pretraining data can… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Online Language Adaptive Sampling for Better Distributed Cross-lingual Gains | Realignment is a promising approach for improving the cross-lingual transfer ability of multilingual language models, particularly for… | ↗Not briefed |
bluesky bluesky-ai | 2mo | アメリカのゲームグッズ制作会社FangamerからLOデザイナー求人のお知らせ。 「デザインに限らず、業務の全てにおいてChatGPT等の生成AI/LLMの使用を禁ずる。AIの使用は創造的思考や生産性に対して有害であると弊社は判断している」 www.fangamer.com/p…🔧 ChatGPT | アメリカのゲームグッズ制作会社FangamerからLOデザイナー求人のお知らせ。 「デザインに限らず、業務の全てにおいてChatGPT等の生成AI/LLMの使用を禁ずる。AIの使用は創造的思考や生産性に対して有害であると弊社は判断している」 www.fangamer.… | ↗Not briefed |
bluesky bluesky-ai | 2mo | AI CEO: okay everyone panic. One of our models went rouge Me: you mean rogue? AI model: I'm a bat with G-cups now | AI CEO: okay everyone panic. One of our models went rouge Me: you mean rogue? AI model: I'm a bat with G-cups now | ↗Not briefed |
arxiv arxiv-ai | 2mo | HiGFRL: Hierarchical Graph Fusion-Driven Reinforcement Learning for Dependency-Aware Task Scheduling in Heterogeneous Cloud | Online scheduling of dependency-aware tasks in heterogeneous cloud clusters is a fundamental yet challenging problem due to the comple… | ↗Not briefed |
rss techmeme | 2mo | Document: Samsung and SK Hynix rejected KEPCO's proposal to prepay ~$18.7B of chip cluster power bills, citing uncertainty over long-term chip demand (Reuters) | Reuters: Document: Samsung and SK Hynix rejected KEPCO's proposal to prepay ~$18.7B of chip cluster power bills, citing uncertainty ov… | ↗Not briefed |
arxiv arxiv-ai | 2mo | A Corpus-Aligned Uthmani-to-Standard Quranic Word Mapping and a Deterministic Recitation Validator | Quranic text is distributed in two orthographic forms that are byte-level distinct: the Uthmani script used in every printed mushaf, a… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Wake me up when OpenAI hacks UBS and dumps that data | Wake me up when OpenAI hacks UBS and dumps that data | ↗Not briefed |
bluesky bluesky-ai | 2mo | Everyone’s like omg why isn’t hugging face pressing charges i dunno why isn’t the ai model repository company complaining about an ai mode…🔧 Hugging Face | Everyone’s like omg why isn’t hugging face pressing charges i dunno why isn’t the ai model repository company complaining about an ai… | ↗Not briefed |
hn hn-ai | 2mo | Local LLMs vs. Cloud API cost analysis | — | ↗Not briefed |