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 | 4d | Subprocess-Constrained Markov Decision Processes | Constrained Markov decision processes (CMDPs) typically bound expected cumulative cost from an initial state. Such a guarantee can be… | ↗Not briefed |
arxiv arxiv-ai | 4d | Variationally correct physics-informed operator learning for nonlinear partial differential equations | Physics-informed operator learning trains neural operators on the residual of a parametric partial differential equation (PDE) instead… | ↗Not briefed |
arxiv arxiv-ai | 4d | Generative Data Augmentation Without External Data: An Empirical and Theoretical Study on Deep Networks | Generative data augmentation adds samples from generative models to a classifier's training set. Many influential studies use generato… | ↗Not briefed |
bluesky bluesky-ai | 4d | You can read the letter we sent to state & local financial officers urging them to stop funding billionaires & to protect working people fr… | You can read the letter we sent to state & local financial officers urging them to stop funding billionaires & to protect working peop… | ↗Not briefed |
hn hn-ai | 4d | Carbon – An experimental Apple Silicon DSL for infernce | — | ↗Not briefed |
hn hn-ai | 4d | Show HN:The Python wheels PyPI is missing | — | ↗Not briefed |
arxiv arxiv-ai | 4d | Warm Starting Neural Dynamics Models From Few Jacobian Samples | Model-based reinforcement learning (MBRL) leverages a model of the environment, used for planning and generating samples without inter… | ↗Not briefed |
bluesky bluesky-ai | 4d | Copyright is so rigidly enforced you can't put a contemporary song lyric in a book without paying thousands in royalties, but anthropic cou… | Copyright is so rigidly enforced you can't put a contemporary song lyric in a book without paying thousands in royalties, but anthropi… | ↗Not briefed |
bluesky bluesky-ai | 4d | ohhh this is huge. available at no cost to approved organizations! and also a @roost.tools shoutout!! deploymentsafety.openai.com/csam-ima… | ohhh this is huge. available at no cost to approved organizations! and also a @roost.tools shoutout!! deploymentsafety.openai.com/csa… | ↗Not briefed |
bluesky bluesky-ai | 4d | “making the thing is annoying/slow” -> “i know! llm, make the thing!” is never gonna work | “making the thing is annoying/slow” -> “i know! llm, make the thing!” is never gonna work | ↗Not briefed |
bluesky bluesky-ai | 4d | The AAUP, along with a coalition of organizations, are calling on banks, institutional investors, & stewards of public funds to: • boycott… | The AAUP, along with a coalition of organizations, are calling on banks, institutional investors, & stewards of public funds to: • bo… | ↗Not briefed |
arxiv arxiv-ai | 4d | Balanced Twin Weight Factorization for Efficient MLLMs | Weight Factorization (WF) methods aim to reduce a model's latency, memory, or both. These typically decompose a pretrained model into… | ↗Not briefed |
bluesky bluesky-ai | 4d | 🤖 OpenAI launches first‑phase embedded assessments with four independent labs OpenAI has signed agreements with four independent assessor…🔧 ChatGPT | 🤖 OpenAI launches first‑phase embedded assessments with four independent labs OpenAI has signed agreements with four independent ass… | ↗Not briefed |
rss techmeme | 4d | Sources: Nvidia's board is not considering bringing Jensen Huang back as CEO and is focused on a national security investigation into Huang (CNBC) | CNBC: Sources: Nvidia's board is not considering bringing Jensen Huang back as CEO and is focused on a national security investigation… | ↗Not briefed |
bluesky bluesky-ai | 4d | i think it would be possible to build a human + llm workflow to keep technical docs updated as code changes (one of the most brutal tasks),… | i think it would be possible to build a human + llm workflow to keep technical docs updated as code changes (one of the most brutal ta… | ↗Not briefed |
arxiv arxiv-ai | 4d | A Scalable Multi Task Learning Framework for Modeling Demographic Disagreements in Annotation Tasks | Annotators from different demographic backgrounds often disagree in systematic ways due to varying lived experiences, cultural context… | ↗Not briefed |
bluesky bluesky-ai | 4d | Like, I'm fine if Project Zero cut their timeline, because they actually put in the work. Anthropic do not and are a useless tax on society | Like, I'm fine if Project Zero cut their timeline, because they actually put in the work. Anthropic do not and are a useless tax on so… | ↗Not briefed |
arxiv arxiv-ai | 4d | Out of Sight, Into Memory: Compact Visual Latents Retain Local Evidence in Ultra-High-Resolution Remote Sensing | Question answering over ultra-high-resolution (UHR) remote sensing scenes poses a tension for multimodal large language models: global… | ↗Not briefed |
arxiv arxiv-ai | 4d | Expertise capture in workplace AI | AI assistants are becoming part of how knowledge workers think and perform, accumulating the prompts, exemplars, memory and interactio… | ↗Not briefed |
bluesky bluesky-ai | 4d | The ostensible justification is "hey other people with models can find the same vulns too," and this would be true if the vulns were real, … | The ostensible justification is "hey other people with models can find the same vulns too," and this would be true if the vulns were r… | ↗Not briefed |
bluesky bluesky-ai | 4d | 🥩 #ai - Artificial Intelligence, RUSSIA ANTROPIC ChatGPT CLAUDE-ai . . . #CyberSec #Signals #WAR🔧 ChatGPT | 🥩 #ai - Artificial Intelligence, RUSSIA ANTROPIC ChatGPT CLAUDE-ai . . . #CyberSec #Signals #WAR | ↗Not briefed |
arxiv arxiv-ai | 4d | Comments on "Deep Fréchet Regression" | We discuss the Journal of the American Statistical Association (JASA) discussion paper "Deep Fréchet Regression" and explore an altern… | ↗Not briefed |
bluesky bluesky-ai | 4d | The galaxy brain socialist solution is to reinstate Jensen and encourage him to sell his best chips to the Chinese so they can make lots of… | The galaxy brain socialist solution is to reinstate Jensen and encourage him to sell his best chips to the Chinese so they can make lo… | ↗Not briefed |
bluesky bluesky-ai | 4d | "Anthropic cuts its 'kick-you-in-the-balls-really-hard' disclosure window from 90 to 30 days." FUCK OFF | "Anthropic cuts its 'kick-you-in-the-balls-really-hard' disclosure window from 90 to 30 days." FUCK OFF | ↗Not briefed |
bluesky bluesky-ai | 4d | Also.. where's my openai settlement? Where's my Google Gemini settlement? Where's my Apple AI settlement? | Also.. where's my openai settlement? Where's my Google Gemini settlement? Where's my Apple AI settlement? | ↗Not briefed |
rss techcrunch-ai | 4d | FBI says Artex AI open-source hacking tool used to breach networks in ‘under three hours’ | The FBI says that the open-source hacking tool was used in cyberattacks targeting a U.S. IT giant and the South Korean financial indus… | ↗Not briefed |
bluesky bluesky-ai | 4d | hey quick question does this mean that a symmetric approach is now fair game within professional standards, like, if someone has a severe A… | hey quick question does this mean that a symmetric approach is now fair game within professional standards, like, if someone has a sev… | ↗Not briefed |
bluesky bluesky-ai | 4d | Both of Schumer’s daughters are AI people. One at Anthropic and one at Meta. | Both of Schumer’s daughters are AI people. One at Anthropic and one at Meta. | ↗Not briefed |
arxiv arxiv-ai | 4d | Replicating and Extending Weight-Space Programming: Frontier LLMs Compile Programs into Weights and Read Input Layers | Researchers at MIT recently reported that GPT-6 Astra, a frontier large language model (LLM), can write the weights of small neural ne… | ↗Not briefed |
bluesky bluesky-ai | 4d | That's "Anthropic cuts the amount of time it gives people to sift through its garbage slop reports to find the few that actually matter bef… | That's "Anthropic cuts the amount of time it gives people to sift through its garbage slop reports to find the few that actually matte… | ↗Not briefed |
bluesky bluesky-ai | 4d | one of my favorite ongoing laughs is the ai bozos going "jokes on you i managed to scrape your site with claude anyway" truly, congratulat… | one of my favorite ongoing laughs is the ai bozos going "jokes on you i managed to scrape your site with claude anyway" truly, congra… | ↗Not briefed |
hn hn-ai | 4d | Show HN: Claude-multi-desktop – several Claude Desktop accounts on one Mac | — | ↗Not briefed |
bluesky bluesky-ai | 4d | Fill the world with AI NPCs. The environment could be, I don’t know, a reference from street view + some sort of world model (and/or aut… | Fill the world with AI NPCs. The environment could be, I don’t know, a reference from street view + some sort of world model (and/o… | ↗Not briefed |
hn hn-ai | 4d | Poland can build the coolest AI data centre and get cheaper natural gas | — | ↗Not briefed |
bluesky bluesky-ai | 4d | Read the model card here: deploymentsafety.openai.com/csam-image-c... - calls for shared, independent benchmarking of CSAM detection syste… | Read the model card here: deploymentsafety.openai.com/csam-image-c... - calls for shared, independent benchmarking of CSAM detection… | ↗Not briefed |
hn hn-ai | 4d | Style guide in, real app out | — | ↗Not briefed |
bluesky bluesky-ai | 4d | pangram isn't an oracle, its 'just' a system with a very low false positive rate. as far as we can all tell. this article gives no reason t… | pangram isn't an oracle, its 'just' a system with a very low false positive rate. as far as we can all tell. this article gives no rea… | ↗Not briefed |
bluesky bluesky-ai | 4d | Anthropic has passed the US government as SpaceX's top customer, per YF. | Anthropic has passed the US government as SpaceX's top customer, per YF. | ↗Not briefed |
arxiv arxiv-ai | 4d | Learning to Reason with Predictive Gains for Generative Recommendation | Reasoning-enhanced generative recommendation (GR) has attracted increasing attention for leveraging the reasoning capabilities of larg… | ↗Not briefed |
arxiv arxiv-ai | 4d | A Reinforcement Learning Method to Navigate SMEFT Operator Space | In particle physics, discrepancies between experimental measurements and Standard Model (SM) predictions are commonly interpreted with… | ↗Not briefed |
bluesky bluesky-ai | 4d | I can't express how important this milestone is. This is the very first time a model card has ever been published for a CSAM Classifier and… | I can't express how important this milestone is. This is the very first time a model card has ever been published for a CSAM Classifie… | ↗Not briefed |
bluesky bluesky-ai | 4d | After Steve died I didn't even follow up on the Anthropic lawsuit mail I got. I knew it would just be a huge hassle and an even deeper leve… | After Steve died I didn't even follow up on the Anthropic lawsuit mail I got. I knew it would just be a huge hassle and an even deeper… | ↗Not briefed |
bluesky bluesky-ai | 4d | i checked to see if i'm blocked yet (i'm not) and wow yep that's bsky staff alright. | i checked to see if i'm blocked yet (i'm not) and wow yep that's bsky staff alright. | ↗Not briefed |
hn hn-ai | 4d | Intel Diamond Rapids Xeon Gets Up to 1.6TB/S of RAM Bandwidth, 1.28GB of Cache | — | ↗Not briefed |
bluesky bluesky-ai | 4d | I think it comes off as pretty AI to me, but, it's not as bad as lots of other AI writing! if they gave an award for the human prompter th… | I think it comes off as pretty AI to me, but, it's not as bad as lots of other AI writing! if they gave an award for the human prompt… | ↗Not briefed |
bluesky bluesky-ai | 4d | So let’s imagine a reviewer who has their own take on the article but also wants to run it through an LLM to see if it picks things up that… | So let’s imagine a reviewer who has their own take on the article but also wants to run it through an LLM to see if it picks things up… | ↗Not briefed |
bluesky bluesky-ai | 4d | If we want proper investigation and model safety evaluation, we need a regulatory framework. Or we need to look closely at the financial ti… | If we want proper investigation and model safety evaluation, we need a regulatory framework. Or we need to look closely at the financi… | ↗Not briefed |
arxiv arxiv-ai | 4d | Variational Monte Carlo for Quantum Excited States via Nested Low-Rank Approximation | The electronic Schrödinger equation encodes a molecule's electronic structure and quantum properties, but solving it scales exponentia… | ↗Not briefed |
arxiv arxiv-ai | 4d | Just sort it! A note on the LP behind rank-one anti-loop edits | Weight-space programming removes the thinking loops of a small reasoning model by a rank-one edit W' = W + u v^T of an MLP output matr… | ↗Not briefed |
bluesky bluesky-ai | 4d | Anthropic cuts its coordinated vulnerability disclosure window from 90 to 30 days for high- and critical-severity issues, in response to AI… | Anthropic cuts its coordinated vulnerability disclosure window from 90 to 30 days for high- and critical-severity issues, in response… | ↗Not briefed |