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 | Why is it that when AI users talk about their latest project it's never "I made this thing, and here's where you can try it", it's always "… | Why is it that when AI users talk about their latest project it's never "I made this thing, and here's where you can try it", it's alw… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rare Gate Disagreements Can Limit Plasticity: When Gradient Flow Mispredicts Finite-Batch SGD | Population gradient flow is a common tool for reasoning about how neural networks adapt, including after pretraining. We show that it… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Chain-of-Thought to Loops: Non-Autoregressive Latent Reasoning via Looped Transformers | Chain-of-thought (CoT) reasoning often improves language-model performance by giving models additional computation before answering. H… | ↗Not briefed |
hn hn-ai | 1mo | Gigabyte W775-V10-L01 Hands-On Bringing Nvidia GB300 Deskside | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Succinctly: - LLMs are based on the external nature of how *other people* talk - using an LLM to produce what should come from your own sel… | Succinctly: - LLMs are based on the external nature of how *other people* talk - using an LLM to produce what should come from your ow… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SAGE: Sink-Aware Guided Emphasis for Visual Grounding in Vision-Language Decoders | Recent large vision-language models (VLMs) pair a visual encoder with a large language model (LLM) and perform well on diverse image-t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Data ends in March 2026, so about 8 months into Claude Code being available and three months before Fable. The bottlenecks in code review h… | Data ends in March 2026, so about 8 months into Claude Code being available and three months before Fable. The bottlenecks in code rev… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GROB: A Multi-Agent Architecture for Public-Trace Investigation of Candidate Agentic Activity | We present GROB, a multi-agent architecture for investigating candidate autonomous-agent activity through public Internet traces when… | ↗Not briefed |
rss mit-tech-review-ai | 1mo | Building a safer path to autonomous industrial AI | Industrial AI is entering a new phase. After decades of predictive analytics and other specialized applications, advances in foundatio… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Exclusive: Three fired OpenAI employees called on the company to work with outside safety auditors and to preserve the ability to monitor i… | Exclusive: Three fired OpenAI employees called on the company to work with outside safety auditors and to preserve the ability to moni… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Who Verifies the Verifier? Co-Evolving Inspectable Graders with Self-Improving Agents🔧 Consensus | We changed the agent: did it actually get better? Every self-improving agent loop answers this hundreds of times, and every answer com… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Speech Captions: Speech-Rewarded Style Planning for Conversational Text-to-Speech | Natural-language style descriptions provide an interpretable interface between large language models (LLMs) and controllable text-to-s… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ProtoSemImage: Image-Valued Prototypes with Deformable Row Alignment for Interpretable Document Classification | Prototypes in classification models are almost always vectors, and a vector has no readable form. This paper asks what happens when a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Feature Space Adaptation for Effortless Gaussian Process Flows | Outside the linear-Gaussian regime, conditional sampling from Gaussian processes (GPs) is challenging. Recent methods such as FlowGP (… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Generative Adversarial Loops | AI research progress can be viewed as the interaction between two processes: benchmark creation and method discovery. Historically, bo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | using a big fuckin swivel arm to move tons of steel or something paired with image intelligence to spot defects especially if it's Outside … | using a big fuckin swivel arm to move tons of steel or something paired with image intelligence to spot defects especially if it's Out… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SpikeSSL: A Universal Spike Inference Framework with Dynamics-Informed State-Space Layers🔧 Modal | Two-photon calcium imaging is a standard tool for recording large neural populations in vivo, yet inferring spikes accurately across t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Zatom-2: Multitask Pretraining on Atomistic Data for Generative Modeling across Domains | Unified atomistic modeling has the potential to accelerate discovery in chemistry, materials science, and biology by bridging data-ric… | ↗Not briefed |
bluesky bluesky-ai | 1mo | new: your AI can publish to Prototir 🚀 tell Claude Code, Codex, Cursor or any MCP client "publish this to Prototir". you approve once, yo… | new: your AI can publish to Prototir 🚀 tell Claude Code, Codex, Cursor or any MCP client "publish this to Prototir". you approve onc… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI Withdraws 3 Math Papers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | So perhaps it’s that heavy LLM use for things which are meant to come from a person can recreate the symptoms of certain psychological diso… | So perhaps it’s that heavy LLM use for things which are meant to come from a person can recreate the symptoms of certain psychological… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I don’t know quite what the right word is but it comes across as some sort of depersonalisation, or dissociation, or alienation. LLM “thi… | I don’t know quite what the right word is but it comes across as some sort of depersonalisation, or dissociation, or alienation. LLM… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI withdraws three of their recent manuscripts | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | HÇ – Hişê Çêkirî (AI, Artificial Intelligence) MZM – Modela Zimanî ya Mezin (LLM, Large Language Model) HGÇ – Hişê Giştî yê Çêkirî (AGI, … | HÇ – Hişê Çêkirî (AI, Artificial Intelligence) MZM – Modela Zimanî ya Mezin (LLM, Large Language Model) HGÇ – Hişê Giştî yê Çêkirî (… | ↗Not briefed |
bluesky bluesky-ai | 1mo | MZM (LLM) Modela Zimanî ya Mezin (MZM) modeleke ku bi ziman dixebite. Heke Kurdî di perwerdekirina wê de cih bigire, ew dikare: bi Kurdî b… | MZM (LLM) Modela Zimanî ya Mezin (MZM) modeleke ku bi ziman dixebite. Heke Kurdî di perwerdekirina wê de cih bigire, ew dikare: bi Ku… | ↗Not briefed |
hn hn-ai | 1mo | Vallic Cloud is live: all the infrastructure, almost none of the ops | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | SAIL: Scientific Agentic Intelligence via a Science-Aware Loop | We introduce SAIL, an open model with 35B total and 3B active parameters for literature research, scientific coding, and multi-step re… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tracing the Thoughts of a Coding Agent Playing ARC-AGI-3: Lessons for Continual Learning | We study how a coding agent learns across a sequence of abstract reasoning tasks. The agent runs on a frozen foundation model inside a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | There‘s a fair amount of whiffle kicking around in this piece, but one para stands out to me: “...It is currently illegal to develop a fro… | There‘s a fair amount of whiffle kicking around in this piece, but one para stands out to me: “...It is currently illegal to develop… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Closed-loop evaluation of LLM agents for embedded software development🔧 GPT-5 | Large language models (LLMs) are increasingly deployed as coding agents that edit files, run builds and tests, inspect execution resul… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MotiveMob: Motivation as Semantic Action for Closed-Loop Human Mobility Generation | Human mobility generation, an important task in urban research, synthesizes trajectory data for urban planning and transportation mana… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Adversarial Cues in Decision Models Used as Judges: The Role of Request Presentation | An answer judge instructed to grade the final commitment should reject an explicitly wrong final value even when an earlier value matc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 1/3 Bobr mi dneska u potoka nadával do darmotrášů. Četl zprávu: OpenAI i Anthropic rozdávaj nejlepší AI úplně zadarmo. Hurá, žejo? Malé ale… | 1/3 Bobr mi dneska u potoka nadával do darmotrášů. Četl zprávu: OpenAI i Anthropic rozdávaj nejlepší AI úplně zadarmo. Hurá, žejo? Mal… | ↗Not briefed |
arxiv arxiv-ai | 1mo | BioBigBird: A Sparse Attention Model for Long-Range Dependency Processing in Biomedical Text | While domain-specific Large Language Models (LLMs) have encoded vast biomedical knowledge, their limited context windows often hinder… | ↗Not briefed |
bluesky bluesky-ai | 1mo | People who generate any of their code with an LLM and submit it to the project are required to disclose what they generated and are blocked… | People who generate any of their code with an LLM and submit it to the project are required to disclose what they generated and are bl… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Great, seeing people lumping .hack fans as "all in on AI" now probably because of the fucking Claude decomp | Great, seeing people lumping .hack fans as "all in on AI" now probably because of the fucking Claude decomp | ↗Not briefed |
arxiv arxiv-ai | 1mo | Why machines will still not rule the world | In our book Why machines will never rule the world [13, 14] we argue that arti- ficial general intelligence is mathematically impossib… | ↗Not briefed |
bluesky bluesky-ai | 1mo | any time i see the “ai model escapes containment” headlines i think “they should lose their engineering licenses, that shit’s out of spec” | any time i see the “ai model escapes containment” headlines i think “they should lose their engineering licenses, that shit’s out of s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Then they do it again. And quite often, they’re getting the LLM to write their replies to other editors while all this is going on. Some… | Then they do it again. And quite often, they’re getting the LLM to write their replies to other editors while all this is going on.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | …and accuse other editors of trying to destroy the project Then, faced with convincing evidence, they admit they DID use an LLM but they *… | …and accuse other editors of trying to destroy the project Then, faced with convincing evidence, they admit they DID use an LLM but t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The wiki page dealing with users breaking the LLM content is grim for how many LLM users start out by flat lying about their use, with elab… | The wiki page dealing with users breaking the LLM content is grim for how many LLM users start out by flat lying about their use, with… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Policy Alignment: New Signals for Membership Auditing in On-Policy Distillation | On-policy distillation (OPD) trains a student model by aligning its policy with a teacher model on trajectories generated by the stude… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Causal-fate dynamics of unrealized influence | Many dynamical systems generate influences whose consequences are not fully exhausted in the realized trajectory at the moment they ar… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Missing Modality-Aware Calibration for Trustworthy Brain Tumor Segmentation | Multimodal brain tumor segmentation typically leverages multiple MRI modalities, yet incomplete modality acquisition is common in clin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rewiring Semantics, Dynamics, and Control: A Simple yet Effective Action-Centric Tri-Stream Transformer | Vision-Language-Action (VLA) models have emerged as a prominent framework for complex robotic manipulation, building on the strong sem… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Refinement as a Service: Algorithmic Predictor Refinement | Prediction aggregation aims to combine information from multiple predictors into a more informative one. We study this question in the… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The irony of implementing LLM coding requirements by software companies who also gutted their QA Department shouldn’t be lost on anybody | The irony of implementing LLM coding requirements by software companies who also gutted their QA Department shouldn’t be lost on anybo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 人々が OpenAI の証明を解読しようとしてるの，ラマヌジャンみがある | 人々が OpenAI の証明を解読しようとしてるの，ラマヌジャンみがある | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cognition-Oriented Emotion Tracing from Causes to Consequences in Real-World Social Scenes | Affective computing has progressed from categorical emotion recognition to open-ended affective analysis with large multimodal models.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Next in Dystopia Today, OpenAI executive surprised to learn his brain has been removed and his cranium stuffed with silicon chips. www.th… | Next in Dystopia Today, OpenAI executive surprised to learn his brain has been removed and his cranium stuffed with silicon chips. w… | ↗Not briefed |