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 | 1mo | CypherTurn: A Multi-Turn Benchmark for Conversational Text-to-Cypher Evaluation and the Autonomy Divergence | Graph databases are increasingly queried through natural language, yet every existing benchmark evaluates isolated single-turn queries… | ↗Not briefed |
hn hn-ai | 1mo | Anthropic IPO prospectus reveals surging costs, $42B 2025 net loss | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Zugleich wurde am Montagabend bekannt, dass OpenAI die für Oktober geplante Veröffentlichung seines KI-Modells der nächsten Generation, GP… | "Zugleich wurde am Montagabend bekannt, dass OpenAI die für Oktober geplante Veröffentlichung seines KI-Modells der nächsten Generatio… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CF-LoRA: Decoupled Factor Aggregation and Adaptation-Aware Client Clustering for Federated LoRA Fine-Tuning | Federated LoRA fine-tuning enables parameter-efficient adaptation of pre-trained models without sharing private data, but suffers from… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Abductive World Modeling via Causal Representation Learning | The central challenge of world modeling is to learn representations that capture how the world evolves. However, existing world models… | ↗Not briefed |
arxiv arxiv-ai | 1mo | REALHOP: Rethinking Multi-Hop Reasoning Evaluation via Behavioral Auditing | Complex questions often require multi-hop reasoning that connects facts distributed across sources or distant regions of a long contex… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SRJudge: Empowering Large Language Models with Selective Reasoning for Fine-Grained Knowledge Concept Tagging | Knowledge concept tagging aims to assign specific concept or topic labels to educational content, which is essential for both educator… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AMU:Admission and Memory Update for Personalized Conversations---Structured Memory with SLM Guided Control | Large language models (LLMs) have become the foundation of personalized assistants, but maintaining persistent user memory across long… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Lemma.work – Open-source AI coworkers your whole team can share | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Repetition, Not Length: Isolating the Counting Failure in Neural Text-to-Speech | Text-to-speech models loop, truncate and lose count on text that repeats a phrase many times. We show that repetition itself is what b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | lmao perfect | lmao perfect | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic besteedt een derde van zijn lijvige prospectus in het kader van zijn komende beursgang om te waarschuwen over de risico’s van AI.… | Anthropic besteedt een derde van zijn lijvige prospectus in het kader van zijn komende beursgang om te waarschuwen over de risico’s va… | ↗Not briefed |
bluesky bluesky-ai | 1mo | PodCastComic #3 Who still nows what to believe when? @calnewport001.bsky.social #AI #LLM @newyorker.com | PodCastComic #3 Who still nows what to believe when? @calnewport001.bsky.social #AI #LLM @newyorker.com | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Rsync.ai – self-hosted CDC and ETL/ELT, source-available (ELv2) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Equally Good, Yet Different: Benchmarking Rashomon sets in AutoML packages | The Rashomon effect describes the existence of multiple near-optimal models that achieve comparable performance while offering fundame… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TaskBridge: Bridging Unsupervised Tabular Anomaly Detection and In-Context Learning via Virtual Tasks | Unsupervised tabular anomaly detection (TAD) aims to identify anomalous rows in tabular data using normal training samples. While conv… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “OpenAI says its model was researching….” dude. Cam. Stop there. The public pays your salary. OpenAI tech can not do “research”. That is no… | “OpenAI says its model was researching….” dude. Cam. Stop there. The public pays your salary. OpenAI tech can not do “research”. That… | ↗Not briefed |
rss techmeme | 1mo | India plans to invest up to $25B in its nascent deep tech industry, as it looks to catch up with the US and China, after investing $11.6B over the past decade (Priyanka Salve/CNBC) | Priyanka Salve / CNBC: India plans to invest up to $25B in its nascent deep tech industry, as it looks to catch up with the US and Chi… | ↗Not briefed |
hn hn-ai | 1mo | How to host and improve the token speed of an LLM | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | 전에 내가 AI의 미래 모습은 기본 모델 위에 각 기업의 고유 데이터를 얹은 커스텀 모델들을 만드는 방향이 될 거라고 했는데, 그렇게 되는 과정. 기업들은 각 기업의 데이터를 '고가 폐쇄형 모델'들에 넘겨주고 싶지도 않고, 다른 회사들이 모두 AI… | 전에 내가 AI의 미래 모습은 기본 모델 위에 각 기업의 고유 데이터를 얹은 커스텀 모델들을 만드는 방향이 될 거라고 했는데, 그렇게 되는 과정. 기업들은 각 기업의 데이터를 '고가 폐쇄형 모델'들에 넘겨주고 싶지도 않고, 다른 회사들이… | ↗Not briefed |
arxiv arxiv-ai | 1mo | JudgeCast: Time Series Forecasting with Experience-Informed Covariate Judgements | Covariate effects vary across contexts and shift over time, requiring forecasters to assess how to use them for each forecasting conte… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Chinese-Jev: Bringing System One Model to Chinese-Language Tasks | System One models such as Jev offer an efficient alternative to generative language models for tasks that require decisions rather tha… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VStress: Correlation-Aware Auditing and Adaptive Budget Allocation for Repeated Verifiers | Repeated verifier calls are useful only when they contribute conditional information. We introduce VStress, an auditable replay contra… | ↗Not briefed |
bluesky bluesky-ai | 1mo | That's Ed He spent a whole day of calling DeepSeek a Chinese fraud because you can't trust Chinese people and the dollar amount was fake...… | That's Ed He spent a whole day of calling DeepSeek a Chinese fraud because you can't trust Chinese people and the dollar amount was fa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Controlled Decoding Attacks on Black-Box LLMs | Manipulating next-token probabilities during generation can bypass the safety alignment of large language models. Existing approaches,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Purlin: Separating Orchestration from the Datapath of Collectives | Distributed inference depends on GPU collective communication that must keep pace with evolving hardware and specialized workloads. Ho… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cool the Sampler, Not the Learner: Sampling Temperature Moves the Staleness Cliff of Importance-Corrected GRPO | Production RL for language models lets the sampler fall behind the learner and repairs the resulting mismatch with a truncated importa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ER-JEPA: Experience Replay Improves Joint-Embedding Predictive Learning in Language Models | Large language models (LLMs) excel at token-level generation but may learn undesirable abstract semantics and lack comprehensive perce… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT, Gemini and Claude, the three leading artificial intelligence assistants, were asked to recommend the best conveyancing solicitor i…🔧 ChatGPT | ChatGPT, Gemini and Claude, the three leading artificial intelligence assistants, were asked to recommend the best conveyancing solici… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scalable Diffusion SBI for Compositional Inference under Simulator Misspecification | Simulation-based inference is challenging when many heterogeneous observations must be composed, hierarchical latent structure must be… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI suspend le lancement d’un nouveau modèle d’IA, jugé insuffisamment contrôlé | OpenAI suspend le lancement d’un nouveau modèle d’IA, jugé insuffisamment contrôlé | ↗Not briefed |
hn hn-ai | 1mo | Post Capitalism | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Fine-Tuning on Self-Generated and Reward-Weighted Data: Learning Dynamics, Convergence Rates, and Benefits of Off-Policyness | We study the learning dynamics of fine-tuning a policy model on self-generated and reward-weighted data, with particular focus on a ge… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Dual-Channel Robust Group-Relative Policy Optimization via Advantage and Sequence-Weight Estimation | Group-relative policy optimization relies on reward-derived advantages and sequence-level likelihood weights, both of which can be sen… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Safe-by-Design Learning via Energy-based Neural Networks | Learning neural-network models of dynamical systems with safety guarantees is a fundamental requirement for their deployment in safety… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SCA: Spatial Credit Assignment for Reinforcement Learning of GUI Agents | GUI agents automate tasks on digital devices by grounding language instructions in visual interfaces. Existing group-relative reinforc… | ↗Not briefed |
arxiv arxiv-ai | 1mo | WeLike2Party! In-Context Motion Transfer for Multi-Human Image Animation | Human image animation aims to transfer motion from a driving video to subjects in a reference image. Despite remarkable progress in vi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CoEM: Empowering Long-Context Reasoning with Commit-on-Evidence Memory | Long-context reasoning is essential for complex and long-horizon tasks, yet the performance of large language models (LLMs) degrades a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VLALight: A Vision-Language-Action Model for Traffic Signal Control | Traffic signal control (TSC) is essential for improving urban mobility and reducing congestion. Although roadside cameras are widely d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ...Except it's worse, because OpenAI didn't 'invent the personal chatbot footprint', and chatbots are a tool for which we already have low-… | ...Except it's worse, because OpenAI didn't 'invent the personal chatbot footprint', and chatbots are a tool for which we already have… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI can extract hitherto hidden meaning from a LLM (i.e. data-mining), but it does not create that meaning. The idea that AI can rapidly imp… | AI can extract hitherto hidden meaning from a LLM (i.e. data-mining), but it does not create that meaning. The idea that AI can rapidl… | ↗Not briefed |
hn hn-ai | 1mo | Anthropic leaders to control company via 'Founder LLC' | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Independent AI Regulation Urgently Needed as Companies Fail to Control Model Misbehavior 🤖 IA: It's not clickbait ✅ 👥 Users: It's not cl… | Independent AI Regulation Urgently Needed as Companies Fail to Control Model Misbehavior 🤖 IA: It's not clickbait ✅ 👥 Users: It's n… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learn from the Gap: Differential-Aware Advantage Pruning with Adaptive Rollout Sampling for GRPO | Recently, Group Relative Policy Optimization (GRPO) and its variants have been developed for policy optimization and demonstrated nota… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Dating the Model: Hidden Dates in System Prompts Affect LLM Evaluation | Reproducibility is essential for scientific research, yet prior work shows that LLM outputs vary with hardware and batching. We identi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | UPDATED: the government has released the OpenAI email sent to Services Australia on Sept. 10 OpenAI fessed its agent managed to get intern… | UPDATED: the government has released the OpenAI email sent to Services Australia on Sept. 10 OpenAI fessed its agent managed to get i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neuro-Symbolic Computer Use: Learning Reusable Policies for Reliable and Efficient Execution | Many computer tasks recur: the same workflow runs many times, with new inputs and from different starting states. Current computer-use… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic: "We're going to wipe out the whole of humanity in 10 years." Capitalism: "Hot-damn! I'll have a piece of that action!" | Anthropic: "We're going to wipe out the whole of humanity in 10 years." Capitalism: "Hot-damn! I'll have a piece of that action!" | ↗Not briefed |
arxiv arxiv-ai | 1mo | State Transport Routing for Short-horizon Adaptation in Multi-horizon Photovoltaic Forecasting | Recent power measurements provide valuable information for photovoltaic(PV) power forecasting, but directly extrapolating short-term t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PrecogUI: Proactive GUI Agents via Pre-cognitive Simulation and Experience Retrieval | Existing reactive Graphical User Interface (GUI) agents often fail in long-horizon, dynamic scenarios, where unexpected disturbances t… | ↗Not briefed |