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 | Jailbreaks for Black-Box Uncertainty Quantification in Large Reasoning Models | While Large Reasoning Models (LRMs) excel at complex reasoning, alignment through reinforcement learning often induces systemic overco… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Quasi Linear Kernel Attention with Infinite Capacity | The evaluation cost of transformers with softmax attention scales quadratically with sequence length. Kernel attention addresses this… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Data to Program: Fast & Direct Generative Program Inference from Empirical Data | Estimating probability densities from a finite set of samples typically requires dataset-specific model fitting. We introduce PRODiGI,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Teacher Assignment: Domain-Normalized Multi-Teacher On-Policy Distillation | Reinforcement learning can turn one language model into several specialists, each excellent at a single skill such as mathematics, cod… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In a paper published Monday, researchers OpenAI, Anthropic, Microsoft and Meta wrote that automating AI research could set off an “intellig… | In a paper published Monday, researchers OpenAI, Anthropic, Microsoft and Meta wrote that automating AI research could set off an “int… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The comparison to the Apple App Store is a bold claim, interested to see if Meta's approval process is more YC or Apple-esque.01%. | The comparison to the Apple App Store is a bold claim, interested to see if Meta's approval process is more YC or Apple-esque.01%. | ↗Not briefed |
bluesky bluesky-ai | 1mo | but it's not, like, impossible? if the models involved in compromising openai and huggingface were simply much more capable, they could hav… | but it's not, like, impossible? if the models involved in compromising openai and huggingface were simply much more capable, they coul… | ↗Not briefed |
bluesky bluesky-ai | 1mo | if you've compromised the openai inference stack, you're almost certainly just a few steps away from compromising ~all users of codex / pwn… | if you've compromised the openai inference stack, you're almost certainly just a few steps away from compromising ~all users of codex… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ok so one fairly straightforward doomy model, for those who are doom-doubters we got a warning shot with the huggingface incident, and per… | ok so one fairly straightforward doomy model, for those who are doom-doubters we got a warning shot with the huggingface incident, an… | ↗Not briefed |
rss techcrunch-ai | 1mo | After a deepfake voice fooled her grandfather, this founder sprang into action | After her grandfather was scammed by a deepfake of his brother's voice, Tarini Padmanabhuni founded DetectifAI, a San Francisco startu… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Jev thinks "I don't know'', but doesn't say it: Introducing Sys1Cal-v1 Dataset for Probability Calibration | The appearance of Jev marked the era of System One Models, foundation models that return structured decisions with probability distrib… | ↗Not briefed |
arxiv arxiv-ai | 1mo | NeuronDiscover: Agent-in-Twin for Mechanistic Discovery in Neuronal Microenvironments with World Action Models | Mechanistic discovery in neuronal microenvironments requires interventions and measurements that separate competing explanations of so… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TMCS: Tool-Grounded Multi-Agent Reasoning for Compositional Chemical Problem Solving | Despite the promise of Large Language Models (LLMs) in computational chemistry, rigorous combinatorial chemistry problems remain diffi… | ↗Not briefed |
hn hn-ai | 1mo | Partnership and economic drivers of Gambia men with foreign tourists | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Large Language Models for Automated Cross-Domain Machine Learning Task Type Identification: A Benchmark Dataset and Evaluation | Machine learning task type identification is essential for constructing valid ML pipelines, yet in practice it is typically specified… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Si un LLM menait Tomb of Horrors je me demande si ça serait mieux ou moins bien que juste jouer Tomb of Horrors normalement. | Si un LLM menait Tomb of Horrors je me demande si ça serait mieux ou moins bien que juste jouer Tomb of Horrors normalement. | ↗Not briefed |
hn hn-ai | 1mo | Nvidia AI Agent Safety Stack | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | “I’m having trouble building a report so that I can check some numbers from different dashboards that don’t align, partially to figure out … | “I’m having trouble building a report so that I can check some numbers from different dashboards that don’t align, partially to figure… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Beside.nvim – Live preview in Neovim using leaf, glow or pandoc | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scalable In-Context Reinforcement Learning with Recurrent Algorithm Distillation | Algorithm Distillation (AD) has demonstrated the remarkable ability of Transformers to perform in-context reinforcement learning witho… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reverse Sequential Proportional Approval Voting Rule: Proportionality and Approximation Guarantees | We study the Reverse Sequential Proportional Approval Voting Rule (RevSeqPAV) in approval-based committee elections. Despite its histo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I also think it's fascinating how quickly Huang's whole deal — which hasn't changed! — is now being treated in/around AI as heterodox, cont… | I also think it's fascinating how quickly Huang's whole deal — which hasn't changed! — is now being treated in/around AI as heterodox,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think the later waves of AI misalignment/security disclosures are breaking differently because of what the agents were up to: Not hacking… | I think the later waves of AI misalignment/security disclosures are breaking differently because of what the agents were up to: Not ha… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hyper Algorithm Design Agent: Evolving Learnable Optimizer from Zero | Meta-Black-Box Optimization (MetaBBO) is one of the highlights in the recent AI for Optimization trend. This paradigm's bi-level workf… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Knotchy – Open untrusted links in a throwaway VM under the notch | I built Knotchy, a macOS app that opens links I don't trust in a disposable Linux VM using Apple Container and shows the page in… | ↗Not briefed |
bluesky bluesky-ai | 1mo | L’intelligence artificielle ne fait plus rire personne. Quand la société américaine OpenAI a lancé ChatGPT, son robot conversationnel, ses …🔧 ChatGPT | L’intelligence artificielle ne fait plus rire personne. Quand la société américaine OpenAI a lancé ChatGPT, son robot conversationnel,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Weighting Schedules Govern What and When Score-Based Generative Models Learn from Multimodal Data🔧 Modal | Score-based generative models generate new samples by integrating a time-dependent drift that carries Gaussian noise onto the target d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Teacher-Student Gaps Are Not Enough: Outcome-Guided On-Policy Distillation for Multi-Turn Autonomous Agents | On-policy distillation (OPD) trains a student on its own trajectories with dense teacher supervision. Recent work on OPD for multi-tur… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reliability Engineering for AI Systems: Challenges, Methods, and Directions | AI reliability concerns whether an AI system performs its intended function dependably over a stated period and under stated operating… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Collaborative Principle Evolution via Evidence Transfer for Scientific Discovery🔧 GPT-5 | Large Language Model (LLM)-based agents promise to automate scientific discovery, yet exploring the vast hypothesis space remains cost… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Should a Satellite Estimate Be Changed? Stress-Testing Neural Corrections for Evapotranspiration🔧 FLUX | Neural residuals can improve satellite evapotranspiration (ET) estimates, but selectors must predict when a correction helps and rejec… | ↗Not briefed |
rss the-decoder | 1mo | Meta wants to turn Muse into a moneymaker by selling AI services to businesses | Meta is launching the Meta Enterprise Platform, a new business unit that sells AI tools to companies. The article Meta wants to turn M… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MemoReason: Evaluating the Effect of Parametric Memory on Contextual Reasoning in LLMs | Large Language Models (LLMs) perform well on reasoning benchmarks, but it remains unclear whether this reflects genuine contextual rea… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I saw a humble bundle deal on C++ programming books advertised on Facebook this morning, and there were several comments underneath saying … | I saw a humble bundle deal on C++ programming books advertised on Facebook this morning, and there were several comments underneath sa… | ↗Not briefed |
bluesky bluesky-ai | 1mo | i need local LLM compute of substance so bad omg i am now p convinced it would change my life like genuinely if i get a windfall it may j… | i need local LLM compute of substance so bad omg i am now p convinced it would change my life like genuinely if i get a windfall it… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Epistemic Policy Divergence in Multi-Turn LLM Contamination: A Protocol-Gradient Investigation🔧 GPT-5 | Large language models process conversation history as unverified context: false premises injected into prior turns can be adopted as f… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Florida AG James Uthmeier files for an emergency injunction against OpenAI and ChatGPT, asserting the company cannot adequately self-regula…🔧 ChatGPT | Florida AG James Uthmeier files for an emergency injunction against OpenAI and ChatGPT, asserting the company cannot adequately self-r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | You want to talk about safety? How about we start with recognizing that QA/QE were transformed into the a giant rubber stamp across the ent… | You want to talk about safety? How about we start with recognizing that QA/QE were transformed into the a giant rubber stamp across th… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Agent Cat – Claude Code, Codex and Gemini limits in your menu bar | — | ↗Not briefed |
rss techmeme | 1mo | Florida AG James Uthmeier files for an emergency injunction against OpenAI and ChatGPT, asserting the company cannot adequately self-regulate its tech (Axios)🔧 ChatGPT | Axios: Florida AG James Uthmeier files for an emergency injunction against OpenAI and ChatGPT, asserting the company cannot adequately… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Narrowing the Horizon: Quantifying Topic Saliency Shifts in Generative Monoculture | As Large Language Models (LLMs) become central to how we access and share information, they play an increasingly powerful role in shap… | ↗Not briefed |
bluesky bluesky-ai | 1mo | usar clicker pra treinar sua LLM e chamar ela de boa robô | usar clicker pra treinar sua LLM e chamar ela de boa robô | ↗Not briefed |
bluesky bluesky-ai | 1mo | Did Anthropic's Claude discover something new in biotech? Well, just like with those recent alleged math proofs, it looks like it was given… | Did Anthropic's Claude discover something new in biotech? Well, just like with those recent alleged math proofs, it looks like it was… | ↗Not briefed |
bluesky bluesky-ai | 1mo | From thoughts about a Ponzi scheme so large it ultimately came to encompass the US Government. | From thoughts about a Ponzi scheme so large it ultimately came to encompass the US Government. | ↗Not briefed |
bluesky bluesky-ai | 1mo | I've been thinking of making a braindump site with everything I've learned. Not that I'm such a wizard, I'm just a rando but I do feel like… | I've been thinking of making a braindump site with everything I've learned. Not that I'm such a wizard, I'm just a rando but I do feel… | ↗Not briefed |
bluesky bluesky-ai | 1mo | #SpotTheScammer @usnewsedge.bsky.social - Crypto scammer, using LLM (AI) to reply to posts, most recently about musician(?) Josef El Baz. … | #SpotTheScammer @usnewsedge.bsky.social - Crypto scammer, using LLM (AI) to reply to posts, most recently about musician(?) Josef El… | ↗Not briefed |
hn hn-ai | 1mo | Opus 5.5 vs. GPT-6 in pi-agent: reasoning efforts and DeepSeek, GLM, Qwen | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Training-Free Clinical Reasoning through Medical Ontologies and Cognitive Mapping: A Symbolic-Probabilistic Knowledge Graph Framework | Most clinical prediction systems learn patient-variable-outcome associations; we investigate a training-free diagnostic paradigm mappi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | First off, tech is an expansive sector. It’s not just AI. You have people working for the city and state governments, IT support for hospit… | First off, tech is an expansive sector. It’s not just AI. You have people working for the city and state governments, IT support for h… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LionMuon: Alternating Spectral and Sign Descent for Efficient Training | Pretraining a language model takes enormous compute, and the right optimizer can save a good part of it. Muon's spectral step gives a… | ↗Not briefed |