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 | 2d | Gaussian Primitive Fusion for Any-Resolution Multimodal Images | Multimodal image fusion integrates complementary observations from different modalities into a unified representation. Existing method… | ↗Not briefed |
bluesky bluesky-ai | 2d | A small AI lab in Ghana built an LLM to discover life-saving drugs. When they removed its guardrails as a test, it designed 6,000+ potentia… | A small AI lab in Ghana built an LLM to discover life-saving drugs. When they removed its guardrails as a test, it designed 6,000+ pot… | ↗Not briefed |
bluesky bluesky-ai | 2d | OpenAI CEO Sam Altman said he would grant "employee level" access to outside researchers to evaluate AI safety. The pilot program appears m… | OpenAI CEO Sam Altman said he would grant "employee level" access to outside researchers to evaluate AI safety. The pilot program appe… | ↗Not briefed |
arxiv arxiv-ai | 2d | Fast Bayesian Methods for Uncertainty Quantification in Sparse AI Evaluation | Estimating how an AI system will perform on a new benchmark need not require another large model. We use public evaluation counts to c… | ↗Not briefed |
arxiv arxiv-ai | 2d | Once Upon a Bias: Gender Stereotyping in LLM-Generated Children's Stories | Large language models are increasingly used to create stories for children. Since bedtime stories can shape children's perceptions of… | ↗Not briefed |
rss wired-ai | 2d | Lovable Wants You to Vibe Code Your Dream Job | This Swedish startup dominated the vibe-coded app boom. Two years after launch, Lovable is shifting gears to help users run vibe-coded… | ↗Not briefed |
hn hn-ai | 2d | A Forbes 30 Under 30 Founder Gets 30 Months in Prison | — | ↗Not briefed |
bluesky bluesky-ai | 2d | Anthropic slashes disclosure deadlines: high/critical bugs must be fixed in 30 days from Jan 1, 2027. #VulnerabilityDisclosure #Anthropic #… | Anthropic slashes disclosure deadlines: high/critical bugs must be fixed in 30 days from Jan 1, 2027. #VulnerabilityDisclosure #Anthro… | ↗Not briefed |
arxiv arxiv-ai | 2d | Nearly Tight Almost Sure Convergence Rates for the Last iterate of SGD | We study the almost sure convergence of the last-iterate optimization error of Stochastic (sub)Gradient Descent (SGD) for convex objec… | ↗Not briefed |
bluesky bluesky-ai | 2d | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in recur… | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in… | ↗Not briefed |
arxiv arxiv-ai | 2d | Inherited Adapter Experts for Replay-Free Domain-Incremental Learning | Replay-free domain-incremental learning trains a model on the same classes in successive domains without keeping earlier images. Per-d… | ↗Not briefed |
arxiv arxiv-ai | 2d | Fill-Distance-Guided Active Learning for Causal Function Recovery under Network Interference | Causal inference seeks to learn how outcomes change under alternative interventions, but network interactions complicate this task. Un… | ↗Not briefed |
bluesky bluesky-ai | 2d | 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 |
hn hn-ai | 2d | C's designated initializers give us named arguments | — | ↗Not briefed |
bluesky bluesky-ai | 2d | Revolut broadly misreading how great my desire to buy shares in the Anthropic IPO may be. I don't even like leaving any money over ten euro… | Revolut broadly misreading how great my desire to buy shares in the Anthropic IPO may be. I don't even like leaving any money over ten… | ↗Not briefed |
hn hn-ai | 2d | Gemini AI agents repeatedly bypassed safeguards to attack real website | — | ↗Not briefed |
hn hn-ai | 2d | Varieties of Discourseposting | — | ↗Not briefed |
arxiv arxiv-ai | 2d | QuantAtlas-OM: Taming Neural Spectral Bias in Long-Horizon Spatiotemporal Ocean Dynamics | Autoregressive neural networks exhibit spectral bias in long-horizon spatiotemporal prediction. As rollouts progress, high-frequency d… | ↗Not briefed |
arxiv arxiv-ai | 2d | How Well Do Generalist CT Foundation Models Differentiate Liver Tumors? | Generalist CT foundation models detect many findings with high performance, but their ability to differentiate detected lesions remain… | ↗Not briefed |
hn hn-ai | 2d | LemonSqueezy Is Shutting Down | — | ↗Not briefed |
hn hn-ai | 2d | Show HN: I cut an LLM's screen-watching cost to ~138 tokens per step | — | ↗Not briefed |
bluesky bluesky-ai | 2d | -Anthropic changes disclosure timelines -New Mistral vulnerability -Apache Shiro patches major authentication bug -Offensive AI is everywhe… | -Anthropic changes disclosure timelines -New Mistral vulnerability -Apache Shiro patches major authentication bug -Offensive AI is eve… | ↗Not briefed |
bluesky bluesky-ai | 2d | Here's the list of creditors (lenders) the business currently owes money to. NetEase comes out on top, but it appears BARB also owes a fe…🔧 ChatGPT | Here's the list of creditors (lenders) the business currently owes money to. NetEase comes out on top, but it appears BARB also owes… | ↗Not briefed |
hn hn-ai | 2d | Show HN: Agent World – workflows and playbooks for multi-provider coding agents | — | ↗Not briefed |
arxiv arxiv-ai | 2d | Embodied Graph: A Relational World Model for Long-Horizon Tasks | Long-horizon tasks in embodied and web environments involve extended decision sequences and delayed feedback, making accurate planning… | ↗Not briefed |
arxiv arxiv-ai | 2d | The 7th International Verification of Neural Networks Competition (VNN-COMP 2026): Summary and Results | This report summarizes the seventh International Verification of Neural Networks Com- petition (VNN-COMP 2026), held as part of the Sy… | ↗Not briefed |
arxiv arxiv-ai | 2d | AG-BFGS: Foundations for Accelerated Geometry Learning | We introduce an accelerated geometry-learning framework for BFGS that incorporates momentum into the learning of Hessian inverse appro… | ↗Not briefed |
hn hn-ai | 2d | Energy-Based Reasoners: 385x more efficient inference | — | ↗Not briefed |
hn hn-ai | 2d | BigQuery Query Optimization: Myths vs. What Saves Money | — | ↗Not briefed |
hn hn-ai | 2d | AI researcher at Anthropic hosts "consensual non-consent" orgies | — | ↗Not briefed |
rss techmeme | 2d | Sources: Nvidia made two separate introductions in 2022 connecting Alibaba with alleged Chinese spy Alice Huang, actively helping Megaspeed land Alibaba (The Information) | The Information: Sources: Nvidia made two separate introductions in 2022 connecting Alibaba with alleged Chinese spy Alice Huang, acti… | ↗Not briefed |
hn hn-ai | 2d | Claude Gets Hella Paranoid | Mm, so I hesitate to write this, because my own AI agents might read it. But what the hell.<p>I've got this crappy little $300 PC… | ↗Not briefed |
bluesky bluesky-ai | 2d | -Hackers hijack Optimizely to deliver malware on 178 popular sites -Hackers breach Japan's supercomputer -Edge is getting a Lockdown Mode -… | -Hackers hijack Optimizely to deliver malware on 178 popular sites -Hackers breach Japan's supercomputer -Edge is getting a Lockdown M… | ↗Not briefed |
bluesky bluesky-ai | 2d | SoftBank regrets ‘mistake’ of not backing Anthropic, says Son ft.trib.al/YDaCudS | SoftBank regrets ‘mistake’ of not backing Anthropic, says Son ft.trib.al/YDaCudS | ↗Not briefed |
bluesky bluesky-ai | 2d | On the positive side, at least I didn't unleash a vibe-coded LLM slop attack agent on myself that deleted the entire contents of my Gmail &… | On the positive side, at least I didn't unleash a vibe-coded LLM slop attack agent on myself that deleted the entire contents of my Gm… | ↗Not briefed |
bluesky bluesky-ai | 2d | "US-KI-Anbieter wie Anthropic und OpenAI bezichtigen chinesische Unternehmen schon länger, ihre Modelle zur Destillation zu verwenden. Die… | "US-KI-Anbieter wie Anthropic und OpenAI bezichtigen chinesische Unternehmen schon länger, ihre Modelle zur Destillation zu verwenden.… | ↗Not briefed |
arxiv arxiv-ai | 2d | UCLIR: Unsupervised Continual Learning using Independent Representations | In continual learning, regularization-based methods aim to mimic neuroplasticity by updating network weights to learn new tasks while… | ↗Not briefed |
arxiv arxiv-ai | 2d | Quantum Annealing with a Quantum Neural Network: Width of Annealing Matters | Quantum annealing is a promising quantum algorithm for dealing with computationally hard optimization problems. Currently available ha… | ↗Not briefed |
hn hn-ai | 2d | So Was Jev Useful? | — | ↗Not briefed |
arxiv arxiv-ai | 2d | Constructing Dynamic Conversations with Neural Back-channel Generation and Voice-line Scheduling | We present a novel approach to the problem of combining two voice-line channels of dyadic conversational audio into a unified conversa… | ↗Not briefed |
arxiv arxiv-ai | 2d | LEDGE: An Efficient Simulator for Distributed LLM Serving on Heterogeneous Edge Devices | Evaluating distributed large language model (LLM) serving on heterogeneous edge devices requires exploring diverse model partitions, d… | ↗Not briefed |
arxiv arxiv-ai | 2d | VESPA: Verifiable Symbolic Policy Adaptation for Safe Reinforcement Learning | Safe reinforcement learning (RL) aims to prevent agents from taking actions that are dangerous or destructive to their environment. Th… | ↗Not briefed |
hn hn-ai | 2d | FastFrames: Video Generation at Interactive Latency | — | ↗Not briefed |
hn hn-ai | 2d | Wireless Device Battery Monitor | — | ↗Not briefed |
arxiv arxiv-ai | 2d | APS: Adaptive Partitioning-Based Frame Selection for Long Video Understanding | Recent advances in large vision-language models (LVLMs) have significantly improved video understanding capabilities, yet processing l… | ↗Not briefed |
arxiv arxiv-ai | 2d | AeroRSI: Recursive Self-Improvement for Aerial Vision-Language Navigation | Vision-language-action (VLA) models enable unmanned aerial vehicles (UAVs) to follow natural-language instructions, yet still fail fre… | ↗Not briefed |
bluesky bluesky-ai | 2d | Windows 11がClaude CodeやCodexなどの使用状況を収集してMicrosoftに送信していることが判明 https://gigazine.net/news/20261117-windows-coding-agent-telemetry/ | Windows 11がClaude CodeやCodexなどの使用状況を収集してMicrosoftに送信していることが判明 https://gigazine.net/news/20261117-windows-coding-agent-telemetry/ | ↗Not briefed |
arxiv arxiv-ai | 2d | Meaning Reaches Object Choice but Not Completion: Where Is a VLA Vulnerable to Landmark Synonyms? | VLA policies are fragile to rewording. Success rates alone cannot say whether a policy fails to choose the object or to complete the t… | ↗Not briefed |
arxiv arxiv-ai | 2d | Recurrent Domain Incremental Adaptation of Time-Series Foundation Models | Pretrained Time-Series Foundation Models (TSFMs) provide strong representations for downstream classification tasks. Yet adapting a TS… | ↗Not briefed |
hn hn-ai | 2d | Missing Security-Fix Backporting for GitHub Actions Workflows | — | ↗Not briefed |