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 | 21d | How Robust Are Prognostic Radiomic Features? Comparing Selection Techniques on Pooled Glioblastoma Cohorts | Radiomic survival studies in glioblastoma report heterogeneous prognostic features, but how much of this stems from the methodology it… | ↗Not briefed |
bluesky bluesky-ai | 21d | Amazon uses AI surveillance to monitor union risk in UK warehouses. Internal documents suggest tech giant collects ‘hundreds of data point… | Amazon uses AI surveillance to monitor union risk in UK warehouses. Internal documents suggest tech giant collects ‘hundreds of data… | ↗Not briefed |
bluesky bluesky-ai | 21d | Great piece here about the Irish government's AI summit and Ireland's economic "model" and identity. | Great piece here about the Irish government's AI summit and Ireland's economic "model" and identity. | ↗Not briefed |
arxiv arxiv-ai | 21d | LeakBench: Auditing Reward Hacking through Prediction-Time Leakage in Clinical Modeling Agents | Autonomous clinical-modeling agents can optimize predictive performance while violating conditions that make a model usable. We introd… | ↗Not briefed |
hn hn-ai | 21d | Share your AI with your friends and team mates | — | ↗Not briefed |
bluesky bluesky-ai | 21d | Actually, now I'm envisaging an AI model called GoLLM, which produces text where every second word is "Precious". Or "Fish" 🐟. | Actually, now I'm envisaging an AI model called GoLLM, which produces text where every second word is "Precious". Or "Fish" 🐟. | ↗Not briefed |
bluesky bluesky-ai | 21d | lobsters goes slop apologist https://lobste.rs/s/h0eate/updates_site_guidelines_ai_llm_tags | lobsters goes slop apologist https://lobste.rs/s/h0eate/updates_site_guidelines_ai_llm_tags | ↗Not briefed |
rss techmeme | 21d | Elon Musk says "I will give one of my SpaceX shares to every kid in America" in 2027; at SpaceX's $146.03 Friday close, the gift would be valued at ~$10.5B (Kai Nicol-Schwarz/CNBC) | Kai Nicol-Schwarz / CNBC: Elon Musk says “I will give one of my SpaceX shares to every kid in America” in 2027; at SpaceX's $146.03 Fr… | ↗Not briefed |
bluesky bluesky-ai | 21d | Apparently they have now fed 'Mean Girls' into the LLM. | Apparently they have now fed 'Mean Girls' into the LLM. | ↗Not briefed |
bluesky bluesky-ai | 21d | Elon Musk pledged to distribute a share of SpaceX to each American child after competitor Anthropic said it would set aside a portion of it… | Elon Musk pledged to distribute a share of SpaceX to each American child after competitor Anthropic said it would set aside a portion… | ↗Not briefed |
arxiv arxiv-ai | 21d | Not All Similarity Is Leakage: Distribution-Aware Identity Leakage Assessment in Synthetic Face Datasets | Synthetic data provides the means for more ethical and less biased facial recognition. Nonetheless, the limitations of generative mode… | ↗Not briefed |
arxiv arxiv-ai | 21d | CUA-Lite: An Open Platform for Computer Use Agents🔧 Replicate | Open resources for computer use agents (CUAs) -- environments, datasets, and models -- remain scattered: without a standard, these res… | ↗Not briefed |
hn hn-ai | 21d | The GPU never reads the same text twice | — | ↗Not briefed |
arxiv arxiv-ai | 21d | FinSafeGuard: A Quality-Driven Synthetic Dataset for Financial AI Safety | Generating high-quality synthetic datasets at scale remains a fundamental challenge for training specialized AI systems in data-scarce… | ↗Not briefed |
bluesky bluesky-ai | 21d | Direkt ein Knaller. | Direkt ein Knaller. | ↗Not briefed |
bluesky bluesky-ai | 21d | Ok merci pour les precision. De base j ai vu sur son compte le plus gros que ca utilise claude etc .du moment ou tu utilise de l ia gen pr … | Ok merci pour les precision. De base j ai vu sur son compte le plus gros que ca utilise claude etc .du moment ou tu utilise de l ia ge… | ↗Not briefed |
bluesky bluesky-ai | 21d | Ah ça je dis pas, je veux juste pas que "generative artist" soit pris pour une génération par ChatGPT ou Midjourney, parce qu'il y a plein …🔧 ChatGPT | Ah ça je dis pas, je veux juste pas que "generative artist" soit pris pour une génération par ChatGPT ou Midjourney, parce qu'il y a p… | ↗Not briefed |
arxiv arxiv-ai | 21d | Honeyed Words, Hidden Daggers: Benchmarking and Training LLMs to See Through Mock Politeness | The Chinese idiom "kou mi fu jian" ("honeyed words, hidden daggers") names speech that is polite in form but hostile in function. We p… | ↗Not briefed |
rss techmeme | 21d | Q&A with Nvidia's Vera team on agentic AI driving CPU demands and forcing CPU design priority back to single-threaded performance over core counts, and more (Wccftech) | Wccftech: Q&A with Nvidia's Vera team on agentic AI driving CPU demands and forcing CPU design priority back to single-threaded perfor… | ↗Not briefed |
hn hn-ai | 21d | PyScript is entering maintenance mod | — | ↗Not briefed |
rss mit-tech-review-ai | 21d | AI researchers stranded Claude on a virtual desert island. It started to “play.” | What do AI agents do if you don’t tell them to do anything? The answer looks a lot like play. Researchers repeatedly stranded them alo… | ↗Not briefed |
arxiv arxiv-ai | 21d | Balance-Aware Pivot Selection for Parallel Diffusion LLM Decoding | Diffusion language models enable parallel token generation, offering a promising alternative to sequential autoregressive decoding. Ex… | ↗Not briefed |
rss wired-ai | 21d | Disney Asked Character.AI to Go Bankrupt for a Get-Out-of-Lawsuits Card🔧 Character.AI | The entertainment giant hopes to get trust and safety tech, among other IP, from a startup whose chatbots dozens of states allege enco… | ↗Not briefed |
arxiv arxiv-ai | 21d | FinHopper: A Progressive Multi-Hop Benchmark for Complex Financial Agents with Step-wise Evaluation | Financial agents are expected to answer factual investment queries through open search, but financial multi-hop retrieval differs from… | ↗Not briefed |
arxiv arxiv-ai | 21d | Decision4Rec: A Jev-Inspired Framework of Decision Models on Knowledge and Active Graphs for Multi-Agent Recommender Systems | Decision models, such as the commercial Jev, are language models that return one probability for each option of a question instead of… | ↗Not briefed |
bluesky bluesky-ai | 21d | Blocked so many game subs because every second post is some idiot going " I decOMPiled it wIth Claude!!1!" or arguments about how AI is act… | Blocked so many game subs because every second post is some idiot going " I decOMPiled it wIth Claude!!1!" or arguments about how AI i… | ↗Not briefed |
bluesky bluesky-ai | 21d | I made an App to detect AI Slop Music. It runs locally on your device and uses Shazam and a custom trained ML model to detect AI generated … | I made an App to detect AI Slop Music. It runs locally on your device and uses Shazam and a custom trained ML model to detect AI gener… | ↗Not briefed |
arxiv arxiv-ai | 21d | AuDQ: Auditable Data Quality through Agentic Rule Discovery | High-stakes tabular pipelines rely on scattered, static quality checks that are hard to evolve and audit against policy intent. Regula… | ↗Not briefed |
arxiv arxiv-ai | 21d | Truncate or Not? A Scaling Study of Looped Transformers with Truncated Backpropagation | Looped transformers reuse a shared block of layers to increase computational depth without increasing the number of trainable paramete… | ↗Not briefed |
arxiv arxiv-ai | 21d | Towards instance-dependent regret optimality in Episodic MDPs with Posterior Sampling | We study regret minimization in finite-horizon episodic Markov Decision Processes (MDPs). While minimax-optimal algorithms are known,… | ↗Not briefed |
arxiv arxiv-ai | 21d | TamedFlow: Trajectory-Aware Distillation\\for Teacher-Assisted Diffusion Sampling | Diffusion Transformers (DiTs) deliver high-fidelity image and video generation, but inference requires many sequential model evaluatio… | ↗Not briefed |
arxiv arxiv-ai | 21d | Multimodal Decoupling for Diversified Recommendation through Semantic and Structural Shaping | Recommender systems often suffer from over-concentrated results when optimized for accuracy, limiting recommendation diversity. While… | ↗Not briefed |
arxiv arxiv-ai | 21d | FrameGRPO: Incentivizing Video Reasoning via Multi-Frame Rollouts and Temporal Credit | Reinforcement learning with verifiable rewards, particularly Group Relative Policy Optimization (GRPO), has improved video reasoning i… | ↗Not briefed |
arxiv arxiv-ai | 21d | Understanding When Reflective Prompt and Skill Optimization Helps on Locally Served Reasoning Models: A Two-Model Study of GEPA and SkillOpt🔧 LM Studio | Reflective text-space optimizers such as GEPA and SkillOpt have reported large gains on frontier and mid-size models. We test replicat… | ↗Not briefed |
hn hn-ai | 21d | Show HN: AI decision model plays Mortal Kombat II | i created a benchmark which evaluates the ai decision models.<p>i thought the idea, by watching jev playing doom<p>i thought, games ar… | ↗Not briefed |
bluesky bluesky-ai | 21d | 日常のひとコマ - ちょっとコンビニへ 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス ➡️ atelier.fan #AIイラスト #ATelier #ATelierで生成 #ブログ #日記 #… | 日常のひとコマ - ちょっとコンビニへ 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス ➡️ atelier.fan #AIイラスト #ATelier #ATelierで生成 #ブログ… | ↗Not briefed |
rss techmeme | 21d | How the data center industry courts kids amid a backlash: 11+ firms and groups hosting field trips since 2025, flag kits and cards for first graders, and more (Caroline Haskins/Wired) | Caroline Haskins / Wired: How the data center industry courts kids amid a backlash: 11+ firms and groups hosting field trips since 202… | ↗Not briefed |
hn hn-ai | 21d | Show HN: A Mojo Script Language Container for the Exasol Database | Mojo is a relatively new language, and its Python affinity makes it an ideal candidate for in-database execution to process data where… | ↗Not briefed |
hn hn-ai | 21d | CacheVerifier, check if verifying LLM cache hits helps your traffic | — | ↗Not briefed |
arxiv arxiv-ai | 21d | ProtGlycanDock: Towards Accurate Protein-Glycan Docking with Tailored Dataset, Benchmark and Model | Performing accurate protein-glycan docking is critical to understanding protein-glycan interactions which govern many biological proce… | ↗Not briefed |
arxiv arxiv-ai | 21d | Accelerated Drift of Scientific Prose in the LLM Era: Evidence from an Interrupted Time-Series Analysis🔧 ChatGPT | Every scientific text is embedded in the writing norms of its time: it reflects earlier publications while also shaping the texts that… | ↗Not briefed |
arxiv arxiv-ai | 21d | Scaling Synthetic Pre-training Data and Evaluation for Indic LLMs | Building capable language models for India requires scalable multilingual data and robust evaluation. We introduce Nemotron-Pretrainin… | ↗Not briefed |
arxiv arxiv-ai | 21d | Scenario Generalization and Sector Transfer in Real Surface Irrigation Surrogate Modelling with Tabular Models | Surface irrigation planning requires screening many operating conditions and spatial configurations, while high-fidelity shallow-water… | ↗Not briefed |
bluesky bluesky-ai | 21d | Det er sørme storladent beskrevet, hva? Har hun stukket hovedet ind i en LLM med nazisprog for at skrive det pis? | Det er sørme storladent beskrevet, hva? Har hun stukket hovedet ind i en LLM med nazisprog for at skrive det pis? | ↗Not briefed |
hn hn-ai | 21d | Pcodec is awesome! 15% smaller and 2x faster than ClickHouse's DoubleDelta | — | ↗Not briefed |
arxiv arxiv-ai | 21d | TimesJev: Thinking Fast on Time Series with System 1 Sufficiency | Large language models (LLMs) have recently shown strong potential for time series reasoning, while models that reason over time series… | ↗Not briefed |
hn hn-ai | 21d | What Makes the Marlboro Red Box Impossible to Mistake | — | ↗Not briefed |
arxiv arxiv-ai | 21d | SPLASH: Contrastive Audit Pressure for Detecting and Localizing Security Hallucinations in Generated Code | Code generated by large language models often passes every functional test while remaining insecure. These security hallucinations are… | ↗Not briefed |
arxiv arxiv-ai | 21d | PnP-CF: A Consensus Flow-Matching Solver for General Inverse Problems🔧 Consensus | Plug-and-Play (PnP) methods using continuous-time deep generative models as learned data distribution priors enable synthesizing high-… | ↗Not briefed |
hn hn-ai | 21d | Codes and you LLM will speak just insert files from zenodo an watch | — | ↗Not briefed |