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 | "USA: Wer »Kl« sagt, der ist für Donald Trump jetzt ein Feind" Erklärt #DumbTrump ChatGPT jetzt den Krieg?🔧 ChatGPT | "USA: Wer »Kl« sagt, der ist für Donald Trump jetzt ein Feind" Erklärt #DumbTrump ChatGPT jetzt den Krieg? | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Liquid Inference – auto-routing to competitive LLM marketplace | Our background is trading systems and exchanges. Liquid Inference applies that to LLM inference. On Liquid Inference, model providers… | ↗Not briefed |
arxiv arxiv-ai | 1mo | X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation🔧 Modal | Audio-visual dataset distillation aims to synthesize a small dataset from a large paired audio-visual dataset, while nearly preserving… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I am begging news outlets to stop gulping down the bullshit that companies are "ponder(ing) machine conscience" of statistical models. Use… | I am begging news outlets to stop gulping down the bullshit that companies are "ponder(ing) machine conscience" of statistical models.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Awwww, do they not want the itty bitty plagiarism machine to get its non-existent feelings hurt? www.bbc.co.uk/news/article... | Awwww, do they not want the itty bitty plagiarism machine to get its non-existent feelings hurt? www.bbc.co.uk/news/article... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Over a year ago I met an Anthropic employee at a conference and at one point she mentioned that one of the things she was working on was wh… | Over a year ago I met an Anthropic employee at a conference and at one point she mentioned that one of the things she was working on w… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Frapp – Open-source game giveaway and deal tracker in React Native | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | you don't need to know a password to break into a computer with LLM integration if you can just p… | you don't need to know a password to break into a computer with LLM integration if you can j… | ↗Not briefed |
hn hn-ai | 1mo | Meatpupptry is far creepier than pure "AI" fakery🔧 ChatGPT | Local pharmacy phoneline is answered by "Caroline" with convincing voice, mannerisms and even background typing noise. You o… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Had a task I'm good at reassigned to the LLM, I know this is happening all over so even trying to find work elsewhere wouldn't necessarily … | Had a task I'm good at reassigned to the LLM, I know this is happening all over so even trying to find work elsewhere wouldn't necessa… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I find the decompiling and clean-room re-implementations of the entire Adobe suite to be very poetic. Couldn't have happened to a better co… | I find the decompiling and clean-room re-implementations of the entire Adobe suite to be very poetic. Couldn't have happened to a bett… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic has added a rule against sustained and needless abusive or cruel behavior toward its models. #AI #Anthropic #Claude #TechPolicy | Anthropic has added a rule against sustained and needless abusive or cruel behavior toward its models. #AI #Anthropic #Claude #Tech… | ↗Not briefed |
hn hn-ai | 1mo | Progress Dashboard for the Clean Room Implementation of Adobe Products | — | ↗Not briefed |
hn hn-ai | 1mo | Does Agentic Misalignment Portend Existential Risk? | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Retractions Revisited: A Replication, Extension, and Mechanistic Evaluation of Linguistic Obfuscation Effects in Fraudulent Science | Prior work suggests fraudulent scientists write their articles with higher rates of linguistic obfuscation than genuine scientists, wi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Horrifying thought: 3-4 years from now where malicious/abliterated LLM models are the new NPM supply chain attack. Think you're getting Dee… | Horrifying thought: 3-4 years from now where malicious/abliterated LLM models are the new NPM supply chain attack. Think you're gettin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | On the Encoding of Calendar Features in Covariate-Informed Time Series Foundation Models | Time series foundation models (TSFMs) increasingly offer covariate support, with calendar information being the most common type of ca… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Not All Dimensions Are Equal: Outlier-Aware Attention Linearization with Grouped Quadratic Kernels | Linearizing softmax attention offers a route to efficient Transformer inference, but preserving pretrained attention remains challengi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | An LLM cannot be sentient. It is not a brain. It's not even a brain simulation. If it tells you you are being mean to it, it is responding … | An LLM cannot be sentient. It is not a brain. It's not even a brain simulation. If it tells you you are being mean to it, it is respon… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This freaks me out. “I think we have reached a point where we need to carefully consider the structure, requirements and value of a graduat… | This freaks me out. “I think we have reached a point where we need to carefully consider the structure, requirements and value of a gr… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT helped write Sophie’s suicide note. When VF asked about it, OpenAI CEO Sam Altman gave an unequivocal answer—and, according to Laur…🔧 ChatGPT | ChatGPT helped write Sophie’s suicide note. When VF asked about it, OpenAI CEO Sam Altman gave an unequivocal answer—and, according to… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Inject a Turn, Not a Token: Learning Where to Redirect LLM Reasoning for Self-Correction | Large language models struggle to reliably identify and correct errors in their own reasoning. Unlike prior training-based approaches… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Step Harder, Not a Step Too Far: The Benefits and Limits of Curriculum Difficulty Progression in GRPO | Curriculum learning typically organizes training samples from easy to hard. In GRPO, medium-difficulty samples are attractive for onli… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI can’t do anything, it’s all fake. The OpenAI math results were in the training data. The coding agents are teleoperated. AI winter now! | AI can’t do anything, it’s all fake. The OpenAI math results were in the training data. The coding agents are teleoperated. AI winter… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Putting "Experience with Claude (AI Psychosis)" on my resume so employers know I keep my skills relevant. | Putting "Experience with Claude (AI Psychosis)" on my resume so employers know I keep my skills relevant. | ↗Not briefed |
bluesky bluesky-ai | 1mo | We kept telling them they live in a fantasy techbro bubble. Dunce hats for all these emotionally stunted anti human wastes of space. Esp… | We kept telling them they live in a fantasy techbro bubble. Dunce hats for all these emotionally stunted anti human wastes of space.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Multi-Agent LLM System for Automated Tuning of Specialized Neural Network Models | Many machine learning approaches are two-level: an expert selects a class of methods, within this class the choice of key components i… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Two things can be true at the same time: 1. This reflects some weird and delusional beliefs about machine sentience on the part of the Anth… | Two things can be true at the same time: 1. This reflects some weird and delusional beliefs about machine sentience on the part of the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Pretraining and Test-Time Training Budgets Matter for ARC: A Controlled Study of VARC and TRM | The Abstraction and Reasoning Corpus (ARC) is increasingly used to evaluate whether neural networks can infer task-specific transforma… | ↗Not briefed |
hn hn-ai | 1mo | Claude was wrong about job impacts | — | ↗Not briefed |
hn hn-ai | 1mo | Fun with low-rank vocab matrices (and a bonus test loss reduction?) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scalable Sparse Autoregressive Variational Inference for Non-Gaussian Spatial Fields | Latent field models are essential for spatial statistics, but the posterior distribution of the latent field, even with a Gaussian pro… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TabReliBench: Evaluating Tabular Foundation Models Beyond Predictive Accuracy | Tabular foundation models (TFMs) are increasingly popular for their strong predictive performance. Many also provide posterior predict… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Critical Look at Power Sampling when Test Time Sharpening LLMs | Reinforcement learning with verifiable rewards (RLVR) has substantially improved LLM reasoning, yet growing evidence suggests it mainl… | ↗Not briefed |
bluesky bluesky-ai | 1mo | they're going to do more to to protect chatbots than you or I | they're going to do more to to protect chatbots than you or I | ↗Not briefed |
arxiv arxiv-ai | 1mo | Electron-Density-Informed Molecular Property Prediction Using Surrogate Functionals | Machine learning methods are increasingly used to predict chemical properties directly from the molecular structure to bypass the ofte… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI corpos are way ahead of you there. But not in a good way. Their information campaigns are especially targeting political, religious, and… | AI corpos are way ahead of you there. But not in a good way. Their information campaigns are especially targeting political, religious… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 🤖🫧💥 www.newscientist.com/article/2592... | 🤖🫧💥 www.newscientist.com/article/2592... | ↗Not briefed |
rss theverge-ai | 1mo | Trump’s attempt to rename AI is looking awfully artificial | President Donald Trump has a knack for turning words against his enemies. His first successful presidential run was built on monikers… | ↗Not briefed |
bluesky bluesky-ai | 1mo | the thing is, being mean to claude is something we can do anywhere bc they scrape the internet, fuck you claude www.techspot.com/news/11415… | the thing is, being mean to claude is something we can do anywhere bc they scrape the internet, fuck you claude www.techspot.com/news/… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SelectArena: A Modern Benchmark for Feature Selection on Real-world Tabular Data | Selecting a subset of features (columns) from a tabular dataset is a ubiquitous challenge in real-world applications. Industry experts… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I just read an article about how Google's AI training model chips were sent into LEO and immediately failed because they didn't consider th… | I just read an article about how Google's AI training model chips were sent into LEO and immediately failed because they didn't consid… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Telling your lockpicking app to pick a lock and then freaking out because it started picking locks while you ignored it is not alien superi… | Telling your lockpicking app to pick a lock and then freaking out because it started picking locks while you ignored it is not alien s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The hugging face attack is primarily proof that OpenAI has incompetent engineers who were too lazy/stupid to oversee their own test environ…🔧 Hugging Face | The hugging face attack is primarily proof that OpenAI has incompetent engineers who were too lazy/stupid to oversee their own test en… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: eBay sold listings APIs and MCP | Get up to 200 eBay sold listings in one request, with displayed prices, condition, shipping, and seller details when available. Built… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Size Independent Constraint Models from Input Data and Positive and Negative Example Solutions | In this paper we describe a Constraint Acquisition Tool that generates size independent constraint programs from input data and positi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | As it happens, I think it *is* morally bad to give an LLM inputs that would be cruel if said to a person, but that’s because it’s bad for t… | As it happens, I think it *is* morally bad to give an LLM inputs that would be cruel if said to a person, but that’s because it’s bad… | ↗Not briefed |
bluesky bluesky-ai | 1mo | lol oh yes that’s why. | lol oh yes that’s why. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Axios reports OpenAI and Anthropic executives are preparing for a potential AI catastrophe that could disrupt banks, electricity, water or … | Axios reports OpenAI and Anthropic executives are preparing for a potential AI catastrophe that could disrupt banks, electricity, wate… | ↗Not briefed |
hn hn-ai | 1mo | Getting old Macromedia Director Games to run on modern Hardware | — | ↗Not briefed |