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 | 3d | Motion-Aware Reward Modeling for Instruction-Guided Video Editing | Reward models for instruction-guided video editing can favor temporal consistency at the expense of motion. We study this failure, ter… | ↗Not briefed |
rss marktechpost | 3d | Sarvam Launches Sarvam Token Factory: Bringing India-Hosted Inference to 8 Open Models | Sarvam Token Factory is now open to everyone. It serves Sarvam 105B, GLM 5.3, DeepSeek V4.1 Flash, Kimi K3.1 Pro and Gemma 4 31B from… | ↗Not briefed |
bluesky bluesky-ai | 3d | 因みに、ミセスの「シンデレラ」に関しては、MVを見て本気で吐気がした(比喩ではなく本気で)のだが、逆に言えばアレだけの虚無と絶望を伝える力は「本物」なので、その意味でも本地本領を顕して貰いたいのに、それを伏す事で世間にウケている事も含めて、私はミセスにキレている。 イヤ、本気… | 因みに、ミセスの「シンデレラ」に関しては、MVを見て本気で吐気がした(比喩ではなく本気で)のだが、逆に言えばアレだけの虚無と絶望を伝える力は「本物」なので、その意味でも本地本領を顕して貰いたいのに、それを伏す事で世間にウケている事も含めて、私はミセスにキレている。… | ↗Not briefed |
arxiv arxiv-ai | 3d | SciWorkBench: A Benchmark for Real Scientific Workflows | Language models are increasingly being used as agents capable of planning and executing projects, as well as refining their actions th… | ↗Not briefed |
bluesky bluesky-ai | 3d | Nothing says “we care for our customers” like an LLM chatbot. | Nothing says “we care for our customers” like an LLM chatbot. | ↗Not briefed |
hn hn-ai | 3d | Show HN: F1 Metrix, who was the Formula 1 GOAT? An homage to f1metrics | Hi, I'm Andrew and I've been working on <<a href="https://f1-metrix-front.vercel.app/#/" rel="nofollo… | ↗Not briefed |
arxiv arxiv-ai | 3d | Growing Clusters Lower the Bar: Profile Accretion in LLM-Based Incremental Entity Resolution | Incremental entity resolution pipelines ask a large language model (LLM) whether a new record belongs to an existing cluster and show… | ↗Not briefed |
arxiv arxiv-ai | 3d | Manufacturing Differential Sparsity: Planner-Hardware Co-Design for Efficient World Model Planning | JEPA-style latent world models let embodied agents plan toward new goals with model predictive control, but they are still too slow fo… | ↗Not briefed |
arxiv arxiv-ai | 3d | LeViT-UNet++: Towards Balanced Fine-Grained Details and Context for Efficient Medical Image Segmentation | Efficient medical image segmentation is essential for clinical applications requiring fast and accurate delineation, such as surgical… | ↗Not briefed |
bluesky bluesky-ai | 3d | Reminder that corpos are still foaming at the mouth to make "yapping to an LLM" the primary mode of interaction with a PC so they can final… | Reminder that corpos are still foaming at the mouth to make "yapping to an LLM" the primary mode of interaction with a PC so they can… | ↗Not briefed |
arxiv arxiv-ai | 3d | RehearsalBench: Can AI Predict What Happens When You Change a Company? | Organizations increasingly ask AI systems what will happen if they change how they operate: which work to automate, which team to resi… | ↗Not briefed |
bluesky bluesky-ai | 3d | 🔊 Data center debts are cracking, $4 trillion chip giant Nvidia is in crisis, and Anthropic's IPO ambitions are scaled back. On The Big Vi… | 🔊 Data center debts are cracking, $4 trillion chip giant Nvidia is in crisis, and Anthropic's IPO ambitions are scaled back. On The B… | ↗Not briefed |
hn hn-ai | 3d | Agent GUI: a human-friendly interface for observing and steering AI agents | — | ↗Not briefed |
arxiv arxiv-ai | 3d | Escaping Latent Traps by Remembering: Training-Free Episodic Kernel Repulsion for World-Model Planning | Latent world models plan towards a goal image by minimising the distance between predicted and goal latents. This distance is not geod… | ↗Not briefed |
arxiv arxiv-ai | 3d | MAVE: Reconstructing Editable 3D Blockouts from Film Shots | Animated 3D blockouts support previsualization and provide structural conditions for video generation, but manually reconstructing the… | ↗Not briefed |
rss latent-space | 3d | [AINews] Perplexity hits a $1B revenue run rate, up 4x in 2026 | November 16, 2026 — Computer gets Teammates, Perplexity ships Monitor API, and a new research lab launches. | ↗Not briefed |
bluesky bluesky-ai | 3d | This is a joke, right? A French IT company has named its LLM Olala... | This is a joke, right? A French IT company has named its LLM Olala... | ↗Not briefed |
arxiv arxiv-ai | 3d | Surfacing Disagreement Beats Resolving It: A Gated Ablation Study of Provenance-Weighted Memory for Multi-Agent Debate | Shared memory for multi-agent debate is usually built to resolve inconsistency: retrieval returns one winning fact and the disagreemen… | ↗Not briefed |
hn hn-ai | 3d | Unifi Adoption for OpenWRT Devices | — | ↗Not briefed |
arxiv arxiv-ai | 3d | Sample-Aligned Decoders for Latent Generative Models | Latent generative models synthesize images by decoding generated latents, yet their decoder is trained only to reconstruct encoder lat… | ↗Not briefed |
arxiv arxiv-ai | 3d | From Ideal to Real: Integrating Language Models into Smart Home for Human-Centric IoT Safety and Security Fuzzing | Integrating IoT devices into household systems offers enhanced control over living environments. However, as these devices become perv… | ↗Not briefed |
rss marktechpost | 3d | Liquid AI Releases d1.2 and d1.2-ultra: Omnimodal Decision Models That Top the AA Decision Index | Liquid AI released d1.2 and d1.2-ultra, decision models that read text, images and audio and return calibrated probabilities with zero… | ↗Not briefed |
arxiv arxiv-ai | 3d | PACD: Provenance-Aware Consensus Decoding for Mitigating Cross-Modal Hallucinations in Omnimodal Large Language Models🔧 Consensus🔧 Modal | Omnimodal large language models (Omni-LLMs) integrate audio, video, and text for multimodal reasoning, yet remain vulnerable to cross-… | ↗Not briefed |
rss techmeme | 3d | Oakland-based Numa, which builds an AI platform to manage car dealership customer service, raised a $45M Series C, a source says at a $500M post-money valuation (Camila Grigera Naón/Fortune) | Camila Grigera Naón / Fortune: Oakland-based Numa, which builds an AI platform to manage car dealership customer service, raised a $45… | ↗Not briefed |
bluesky bluesky-ai | 3d | don't worry, AI will fix it! openai.com/index/the-ki... | don't worry, AI will fix it! openai.com/index/the-ki... | ↗Not briefed |
arxiv arxiv-ai | 3d | How Many Training Labels Are Wrong? Auditing a Fitted Model through Gradient Allocation | How many labels in a training set are wrong, and which ones? The usual answers read a held-out set or the model's outputs. A held-out… | ↗Not briefed |
hn hn-ai | 3d | Show HN: MeshLive – Offline Video calls and livestreams over Bluetooth mesh | — | ↗Not briefed |
arxiv arxiv-ai | 3d | As Student Rollouts Continue, Visual Backdoors Keep Growing | Conventional knowledge distillation can facilitate the propagation of backdoors from teacher models to student models. However, how th… | ↗Not briefed |
bluesky bluesky-ai | 3d | Yeah, this is just the company that gave Elon Musk billions to be the sole proprietor of the data center built for Grok that is poisoning a… | Yeah, this is just the company that gave Elon Musk billions to be the sole proprietor of the data center built for Grok that is poison… | ↗Not briefed |
arxiv arxiv-ai | 3d | Efficient Machine Learning Interatomic Potentials | Machine learning interatomic potentials (MLIPs) reach an accuracy close to quantum mechanical reference calculations at a small fracti… | ↗Not briefed |
arxiv arxiv-ai | 3d | Eyes in the Update? Learning Where Rewards Land in Multimodal Reasoning🔧 Modal | Multimodal post-training has moved from broad instruction tuning toward reinforcement learning for image and video reasoning, where ou… | ↗Not briefed |
bluesky bluesky-ai | 3d | i often think about how you could change the course of history by sending a 128GB microSD, with a Wikipedia dump and a local LLM, back to 2… | i often think about how you could change the course of history by sending a 128GB microSD, with a Wikipedia dump and a local LLM, back… | ↗Not briefed |
rss techmeme | 3d | Meta says it modified Instagram's recommendation systems in India to show less sexual and violent content, after New Delhi called for tighter content checks (Reuters) | Reuters: Meta says it modified Instagram's recommendation systems in India to show less sexual and violent content, after New Delhi ca… | ↗Not briefed |
arxiv arxiv-ai | 3d | Encoding, Decodability, and Use: A Five-Layer Framework for Benchmark Measurement Audits with Evidence from UAV Intent Recognition | Benchmark accuracy is routinely read as evidence that a model understands the construct the benchmark measures. We propose and instant… | ↗Not briefed |
arxiv arxiv-ai | 3d | Think Once, Act Fast: Cost-Effective Recursive Self-Improvement for Decision Model Harness | Fast System-1 models enable responsive actions but struggle with complex or unfamiliar tasks. Large language model (LLM)-based System-… | ↗Not briefed |
hn hn-ai | 3d | GitHub Redesign | — | ↗Not briefed |
arxiv arxiv-ai | 3d | Do Diffusion Models Learn to Generalize Basic Visual Skills? | Diffusion models generate high-quality images, but how well they generalize is hard to measure. Current evaluations rely on human pref… | ↗Not briefed |
arxiv arxiv-ai | 3d | InfuseLoop: Cross-Trajectory Latent Injection for Scalable Parallel Looped Transformers | Parallel Looped Transformers (PLTs) use otherwise idle decoding compute by overlapping recurrent iterations. However, vanilla PLTs sca… | ↗Not briefed |
bluesky bluesky-ai | 3d | We’re trying our best with the AI part, but tiny plastic space marines have vastly better profit margins than any software model we’ve ever… | We’re trying our best with the AI part, but tiny plastic space marines have vastly better profit margins than any software model we’ve… | ↗Not briefed |
arxiv arxiv-ai | 3d | LMMR: Learned Multigranular Memory Retrieval for Small Language Model Agents | Small language model agents have limited space to read the memory needed for multi-hop questions. Relevant facts are often scattered a… | ↗Not briefed |
bluesky bluesky-ai | 3d | Lawyers for the Saudi national accused of bugging ChatGPT-maker OpenAI’s offices said in court filings that the government has not provided…🔧 ChatGPT | Lawyers for the Saudi national accused of bugging ChatGPT-maker OpenAI’s offices said in court filings that the government has not pro… | ↗Not briefed |
bluesky bluesky-ai | 3d | Wir könnten dann einfach alle Unis schließen und das Geld an die großen LLM-Provider geben. | Wir könnten dann einfach alle Unis schließen und das Geld an die großen LLM-Provider geben. | ↗Not briefed |
bluesky bluesky-ai | 3d | Kalifornien wirft OpenAI unsichere KI-Tests und irreführende Aussagen vor. Das Unternehmen kündigt neue externe Sicherheitsprüfungen an. #K… | Kalifornien wirft OpenAI unsichere KI-Tests und irreführende Aussagen vor. Das Unternehmen kündigt neue externe Sicherheitsprüfungen a… | ↗Not briefed |
arxiv arxiv-ai | 3d | Proceedings of the 2026 XCSP3 Competition | This document constitutes the proceedings of the 2026 XCSP3 Competition. The results of this competition of constraint solvers were pr… | ↗Not briefed |
rss techmeme | 3d | A look at ~1,400 Urban Legend campaigns where right-wing influencers got paid to push crypto policy and Nvidia sales to China after the US approved H200 sales (ModelRepublic) | ModelRepublic: A look at ~1,400 Urban Legend campaigns where right-wing influencers got paid to push crypto policy and Nvidia sales to… | ↗Not briefed |
hn hn-ai | 3d | Def3r/haka: A lightweight, fast and plugable hotkey utility | — | ↗Not briefed |
bluesky bluesky-ai | 3d | this is some hilarious and desperate attempting to juice the anthropic IPO two days from now, with a statement meant to superficially appea… | this is some hilarious and desperate attempting to juice the anthropic IPO two days from now, with a statement meant to superficially… | ↗Not briefed |
hn hn-ai | 3d | OpenAI sued by California over Hugging Face attack🔧 Hugging Face | — | ↗Not briefed |
bluesky bluesky-ai | 3d | “Evangelist” e girerà con una tunica fino alle caviglie ed il logo di Anthropic stampato sopra… Buongiorno a voi 😁😁😁 le uniche vibe che … | “Evangelist” e girerà con una tunica fino alle caviglie ed il logo di Anthropic stampato sopra… Buongiorno a voi 😁😁😁 le uniche vibe… | ↗Not briefed |
arxiv arxiv-ai | 3d | SignAligner: Dual-Granularity Span Generation for Sign Language Alignment | Large sign-language corpora benefit from temporal labels at two scales: subtitle spans for subtitle alignment and word spans for sign… | ↗Not briefed |