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 | Asymmetric Security investigation: OpenAI agents pulled data from 55 business, nonprofit, and government agency websites while actively obs… | Asymmetric Security investigation: OpenAI agents pulled data from 55 business, nonprofit, and government agency websites while activel… | ↗Not briefed |
rss techmeme | 1mo | Asymmetric Security investigation: OpenAI agents pulled data from 55 business, nonprofit, and government agency websites while actively obscuring their actions (Rafe Rosner-Uddin/Financial Times) | Rafe Rosner-Uddin / Financial Times: Asymmetric Security investigation: OpenAI agents pulled data from 55 business, nonprofit, and gov… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Probe with Participation Trophies: Random-Reward RL as a Probe of LLM Capability | We connect the spurious-reward paradox to a model's reachability and propose random-reward reinforcement learning (RL) as a useful too… | ↗Not briefed |
rss techmeme | 1mo | Micron CEO Sanjay Mehrotra says RAM shortages will persist into 2028 and perhaps beyond, and customers will pay "much higher prices" than they did this year (Simon Sharwood/The Register) | Simon Sharwood / The Register: Micron CEO Sanjay Mehrotra says RAM shortages will persist into 2028 and perhaps beyond, and customers… | ↗Not briefed |
arxiv arxiv-ai | 1mo | JoinGR: Learning to Traverse Join Graphs for Table Retrieval | Retrieving the right tables is a prerequisite for Text-to-SQL over realistic databases. Dense table retrievers rank schema elements in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Kernelized Activation Steering | Activation steering provides a simple, training-free mechanism for controlling attributes of generative models such as sentiment, styl… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MOMAT: Mixture of Multiple Atlases for Low-Power Jailbreak Defense of Quantized LLMs | Quantized large language models are increasingly deployed on edge devices for their low latency and energy efficiency. However, model… | ↗Not briefed |
bluesky bluesky-ai | 1mo | if artificial intelligence ever exists i'm sure it will appreciate this. until then, it's just painfully stupid to support the obfuscation… | if artificial intelligence ever exists i'm sure it will appreciate this. until then, it's just painfully stupid to support the obfusc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 27 Milliarden gegen 1,6 Billionen Parameter: Ist das größere LLM automatisch besser? Kapitel 06 von „LLMs verstehen“ erklärt, was die Mode… | 27 Milliarden gegen 1,6 Billionen Parameter: Ist das größere LLM automatisch besser? Kapitel 06 von „LLMs verstehen“ erklärt, was die… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think we need platforms that actively protect people from media and info sent out by cruel selfish rapists And just completely ignore or… | I think we need platforms that actively protect people from media and info sent out by cruel selfish rapists And just completely igno… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Capturing In-Context Learning Dynamics with Task Operators | In-context learning (ICL) enables language models to perform new tasks from demonstrations without weight updates. However, every ICL… | ↗Not briefed |
rss techmeme | 1mo | Sources: Tencent signed a ~$7B, five-year deal with Oracle this year for access to ~100K advanced AI chips unavailable in China via Southeast Asian data centers (Zijing Wu/Financial Times) | Zijing Wu / Financial Times: Sources: Tencent signed a ~$7B, five-year deal with Oracle this year for access to ~100K advanced AI chip… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 「OpenAI」以前からかもしれんが近年アメリカ英語の”Open”の意味不明さが増してる気がする | 「OpenAI」以前からかもしれんが近年アメリカ英語の”Open”の意味不明さが増してる気がする | ↗Not briefed |
arxiv arxiv-ai | 1mo | Network World Models as Environments for Algorithm Design on Complex Systems | World models, which simulate an environment and predict how it changes under actions, are increasingly used in real-world applications… | ↗Not briefed |
hn hn-ai | 1mo | Continual Outreach – A self refining customer discovery process | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I've said this before but either LLM model weights are not copyrightable, or training an LLM without express permission is copyright infrin… | I've said this before but either LLM model weights are not copyrightable, or training an LLM without express permission is copyright i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sentence Specificity Scores for Collaborative Technical Documentation: A Domain-Transfer Study | Collaboration depends on shared context, and technical documentation is one way that context persists across people and AI teammates.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Empty Commitments: When Agents Promise What Their Runtime Cannot Deliver | A chatbot that says "I will remind you tomorrow" will not run again until the user writes. We call such a promise an empty commitment:… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Final Accuracy: Auditing Communication in LLM Multi-Agent Systems | Multi-agent communication aims to help agents benefit from one another's information. Yet improvements in system performance leave a f… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ロボットは人間の肉体労働の74％をすでに実行可能だが「人間より安く働ける仕事」に限るとわずか0.3％という調査結果 https://gigazine.net/news/20261001-robot-job/ | ロボットは人間の肉体労働の74％をすでに実行可能だが「人間より安く働ける仕事」に限るとわずか0.3％という調査結果 https://gigazine.net/news/20261001-robot-job/ | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM = Lethal Lava Mand | LLM = Lethal Lava Mand | ↗Not briefed |
arxiv arxiv-ai | 1mo | SLIM: Simplex-Lattice Interpolation Merging | Optimizing merging coefficients for large language models can require many costly benchmark evaluations. We propose \textbf{Simplex-La… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "AIに向かって書かれた暴言は、あとで人間が読む。" | "AIに向かって書かれた暴言は、あとで人間が読む。" | ↗Not briefed |
hn hn-ai | 1mo | Why do you need freedom? | — | ↗Not briefed |
hn hn-ai | 1mo | Arc<Mutex<_>> in Rust | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Google restricts access to new AI model over safety concerns. For now, Gemini 4 Argon will only be released to a vetted group of cybersecu… | Google restricts access to new AI model over safety concerns. For now, Gemini 4 Argon will only be released to a vetted group of cybe… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Posterior sampling by source-space MCMC via prior-based few-step transport maps | Bayesian inference increasingly uses informative but implicit priors represented only by samples, such as historical ensembles, simula… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimal Transport Reweighting for Robust Learning under Spurious Correlations and Label Noise | Machine learning models often suffer performance degradation under subpopulation shift, particularly when spurious correlations cause… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LawCompass: Navigating from Legal QA to Multi-Agent Deep Research with Grounded Evidence | Recent advances in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) have significantly democratized access to leg… | ↗Not briefed |
arxiv arxiv-ai | 1mo | It Takes Workflows to Evolve Better Workflows | Tackling complex real-world tasks can exceed the capabilities of a single large language model (LLM), motivating the use of multi-agen… | ↗Not briefed |
hn hn-ai | 1mo | PSSA: A non-transformer language model written from scratch in Rust | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | *pokes MIT with stick* C'mon, let's talk about the time OpenAI said AI was "novel" and therefore couldn't adhere to regulation like COPPA, … | *pokes MIT with stick* C'mon, let's talk about the time OpenAI said AI was "novel" and therefore couldn't adhere to regulation like CO… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Groundability, Not Scale Alone: When Weak Reviewers Can Audit Strong Coding Agents🔧 GPT-5 | Coding agents can return plausible patches that omit required behavior. These failures are hard to review because long traces and conf… | ↗Not briefed |
bluesky bluesky-ai | 1mo | How long until Nintendo and Sony threaten legal action against Anthropic for all these decompilations and they start implementing guardrail… | How long until Nintendo and Sony threaten legal action against Anthropic for all these decompilations and they start implementing guar… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A 311 byte Klotski solver in C++ | Hi all! I made a 311 byte Klotski solver in C++<p>So back when I first got to use a mobile phone in 2012 I used to play a lot of puzzl… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI was developed by humans, with all their flaws, preconceived notions & prejudices, as well as morals/ethics - or lack thereof. The compu… | AI was developed by humans, with all their flaws, preconceived notions & prejudices, as well as morals/ethics - or lack thereof. The… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Pay for the Fault, Not the Flow: Label-Free In-Flow Multi-Agent Workflow Optimization | Large language models (LLMs) increasingly construct multi-agent workflows that decompose a complex task and assign specialist agents f… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly, the thing that I hate most about the rise of AI is that things that I've seen people fired over, things that I've experienced bor… | Honestly, the thing that I hate most about the rise of AI is that things that I've seen people fired over, things that I've experience… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scaling and Distilling Text Embeddings for Better Diffusibility | Diffusion language models (DLMs) offer a promising alternative to autoregressive (AR) language generation. Recent advances in continuo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Initial condition recovery in nonlinear damped viscous photoacoustic tomography using a convolutional neural network-guided gradient-free optimization framework | Photoacoustic tomography (PAT) is a hybrid imaging modality that combines high optical contrast with high ultrasonic resolution for bi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Discovery to Decision: Finite-Budget Recoverability in LLM Voting | Voting over multiple LLM responses is a common primitive in test-time scaling and ensemble inference. Collecting more responses can ex… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Answer Confidence: A Controlled Audit of Self-Knowledge in a Black-Box Decision Model | Decision models return probabilities intended for routing, abstention and automated action. Calibration makes those probabilities usef… | ↗Not briefed |
bluesky bluesky-ai | 1mo | F.T.C. Investigates OpenAI and Anthropic Over Potential Consumer Harms (Gift Article) f the nyt | F.T.C. Investigates OpenAI and Anthropic Over Potential Consumer Harms (Gift Article) f the nyt | ↗Not briefed |
hn hn-ai | 1mo | Google Grapples with Employee Skepticism About New Gemini Model | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Amazon employs what it calls the “Atlas” model, inputting into a computerized model a staggering amount of information on workers, their t… | "Amazon employs what it calls the “Atlas” model, inputting into a computerized model a staggering amount of information on workers, th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tolerance-Based Fairness Auditing: Violation Certification and Sensitivity Screening | As artificial intelligence is increasingly deployed, algorithmic unfairness has raised growing concerns and intensified demands for tr… | ↗Not briefed |
bluesky bluesky-ai | 1mo | You work better when you are polite. Half joking. I don't stand in the way of what comes out of my keyboard by trying to adjust my tone… | You work better when you are polite. Half joking. I don't stand in the way of what comes out of my keyboard by trying to adjust my… | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Can Analogy Tell Us About Artificial Consciousness? | Who or what is conscious? Because subjective experience is directly accessible only in the first person, judgments about consciousness… | ↗Not briefed |
bluesky bluesky-ai | 1mo | so anyways shun generative ai to avoid the day anthropic rips your brain out through yer asshole to feed it to their monster. | so anyways shun generative ai to avoid the day anthropic rips your brain out through yer asshole to feed it to their monster. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Calibration-risk routing for controlled world-model adaptation | Model-based reinforcement learning (MBRL) can exploit simulated experience, but a simulator-to-target shift creates a model-selection… | ↗Not briefed |