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 | 10d | OverServe: Resource Amplification Attacks on SaaS Agents via Tool--Prompt Co-Design | Advances in large language models (LLMs) are enabling software-as-a-service (SaaS) providers to automate longer business workflows wit… | ↗Not briefed |
bluesky bluesky-ai | 10d | Many of the biggest indies out there still get away with using LLM translations, most infamously Silksong's Chinese translation (before the… | Many of the biggest indies out there still get away with using LLM translations, most infamously Silksong's Chinese translation (befor… | ↗Not briefed |
bluesky bluesky-ai | 10d | dogshit take and also normalization of ai usage. foh. | dogshit take and also normalization of ai usage. foh. | ↗Not briefed |
hn hn-ai | 10d | Show HN: Python suite for spectral geometry analysis of 1.5M+ MPCORB orbits | — | ↗Not briefed |
bluesky bluesky-ai | 10d | So now I have to fill in and sign an IRS form as a "beneficial owner" to stand a hope of seeing my money from Anthropic? It's only my utte… | So now I have to fill in and sign an IRS form as a "beneficial owner" to stand a hope of seeing my money from Anthropic? It's only my… | ↗Not briefed |
bluesky bluesky-ai | 10d | my practical response to this question was to just switch to duckduckgo and turn off all their LLM features | my practical response to this question was to just switch to duckduckgo and turn off all their LLM features | ↗Not briefed |
arxiv arxiv-ai | 10d | GenPFN: An In-Context Tabular Foundation Model Approach for Augmenting Small Datasets in Engineering | Costly simulations and experiments limit the data available for engineering design. This paper proposes GenPFN, a training-free candid… | ↗Not briefed |
bluesky bluesky-ai | 10d | People laughing at me when I said the big play in academia with AI was gonna be forcing you to access data through Claude/ChatGPT and not k…🔧 ChatGPT | People laughing at me when I said the big play in academia with AI was gonna be forcing you to access data through Claude/ChatGPT and… | ↗Not briefed |
hn hn-ai | 10d | LLM inference, visually explained: why a 13B model reads 26 GB per token | — | ↗Not briefed |
bluesky bluesky-ai | 10d | eh, some popular friendslop game with no AI disclosure was bullied into admitting to what I would interpret as "mild" AI usage- according t… | eh, some popular friendslop game with no AI disclosure was bullied into admitting to what I would interpret as "mild" AI usage- accord… | ↗Not briefed |
bluesky bluesky-ai | 10d | Anthropic's seven co-founders could each be worth about $18 billion each after the company's IPO. | Anthropic's seven co-founders could each be worth about $18 billion each after the company's IPO. | ↗Not briefed |
rss techmeme | 10d | Coframe, which uses AI to make websites that test and personalize themselves for each visitor, raised a $25M Series A, bringing its total funding to $35M (Ben Bergman/Business Insider) | Ben Bergman / Business Insider: Coframe, which uses AI to make websites that test and personalize themselves for each visitor, raised… | ↗Not briefed |
bluesky bluesky-ai | 10d | Arguments like "The right play was to hire a consultant or contractor or a graphics programmer earlier in the project with a ton of Unity e… | Arguments like "The right play was to hire a consultant or contractor or a graphics programmer earlier in the project with a ton of Un… | ↗Not briefed |
hn hn-ai | 10d | The Surprising Effectiveness of Deleting Weights in LLM Post-Training | — | ↗Not briefed |
bluesky bluesky-ai | 10d | also, rich all the ZERO TOLERANCE WE SHOULD BURN THEM TO THE GROUND THEY ARE EVIL INCARNATE people saying those things on... bluesky, the v… | also, rich all the ZERO TOLERANCE WE SHOULD BURN THEM TO THE GROUND THEY ARE EVIL INCARNATE people saying those things on... bluesky,… | ↗Not briefed |
bluesky bluesky-ai | 10d | Claude chatbot developer had been reported to be aiming for $2tn IPO on Wall Street before investors cooled on AI | Claude chatbot developer had been reported to be aiming for $2tn IPO on Wall Street before investors cooled on AI | ↗Not briefed |
bluesky bluesky-ai | 10d | by insinuating that the issue with ai is the technology itself, you are helping deflect criticism away from openai and claude and google an… | by insinuating that the issue with ai is the technology itself, you are helping deflect criticism away from openai and claude and goog… | ↗Not briefed |
bluesky bluesky-ai | 10d | if you are complaining about the 4 lines of code they copy pasted into big walk from an llm i do not care you are a fed | if you are complaining about the 4 lines of code they copy pasted into big walk from an llm i do not care you are a fed | ↗Not briefed |
bluesky bluesky-ai | 10d | House House used an LLM at the end of production of Big Walk to make a small fix and the worst end of the anti-AI crusaders are acting like… | House House used an LLM at the end of production of Big Walk to make a small fix and the worst end of the anti-AI crusaders are acting… | ↗Not briefed |
hn hn-ai | 10d | How We Built TLX: A Hardware-Native GPU Compiler for Production ML at Scale | — | ↗Not briefed |
bluesky bluesky-ai | 10d | Completely agree with your sentiment here but, to be completely fair, it wasn't just once. Their statement mentions working through multipl… | Completely agree with your sentiment here but, to be completely fair, it wasn't just once. Their statement mentions working through mu… | ↗Not briefed |
arxiv arxiv-ai | 10d | Distributionally Robust BEST-RQ For Speech Representation Learning | BEST-RQ, based on random projection and quantization, has been a well-established self-supervised learning approach to learn speech re… | ↗Not briefed |
bluesky bluesky-ai | 10d | FUCK. OFF. | FUCK. OFF. | ↗Not briefed |
hn hn-ai | 10d | The answer shouldn't depend on which GPU you rented | — | ↗Not briefed |
hn hn-ai | 10d | A simple Python implementation of the ReAct pattern for LLMs | — | ↗Not briefed |
bluesky bluesky-ai | 10d | Before it becomes a crisis: please consider tossing a few $/£/¥/€ to Wikimedia Foundation when you can. Despite Wikipedia's flaws, it's one… | Before it becomes a crisis: please consider tossing a few $/£/¥/€ to Wikimedia Foundation when you can. Despite Wikipedia's flaws, it'… | ↗Not briefed |
bluesky bluesky-ai | 10d | 実は、プロンプトって1行変えるだけで全然結果違うんですよね。正直、これ地味にすごいかも。 One small change in a prompt can seriously alter AI output. It's surprisingly powerful. #AI …🔧 ChatGPT | 実は、プロンプトって1行変えるだけで全然結果違うんですよね。正直、これ地味にすごいかも。 One small change in a prompt can seriously alter AI output. It's surprisingly powerful.… | ↗Not briefed |
arxiv arxiv-ai | 10d | When Is Proof-Step Verification Worth Its Cost? Operating Requirements of Certify-or-Decline Verification for Open-Weight Mathematical Reasoning | Proof-step verification rewrites a solver's answer into typed steps and uses LLM judges to certify or decline. We evaluate this pipeli… | ↗Not briefed |
hn hn-ai | 10d | Show HN: Warmctl – Fast cold starts for vLLM inference🔧 vLLM | — | ↗Not briefed |
bluesky bluesky-ai | 10d | when we're in a world where LLMs are uncovering decades-old vulnerabilities in software I'm not sure that we can really stick to the "it's … | when we're in a world where LLMs are uncovering decades-old vulnerabilities in software I'm not sure that we can really stick to the "… | ↗Not briefed |
hn hn-ai | 10d | Show HN: A Git friendly open API client | Hello HN!<p>My team and I have been using Postman, then moved on to Bruno but we still had issues with how it handled environments, wo… | ↗Not briefed |
bluesky bluesky-ai | 10d | especially when there's much worse studios championing LLM usage like Epic we should be burning down. Yes, make it poison to use it any cap… | especially when there's much worse studios championing LLM usage like Epic we should be burning down. Yes, make it poison to use it an… | ↗Not briefed |
hn hn-ai | 10d | GitLab patch release 19.5.4 fixes CVE-2026-105437 (vulnerability in Git) | — | ↗Not briefed |
bluesky bluesky-ai | 10d | ┈🌹━━━━━━━━🌹┈ ✨ [ #perfectgoddess] ✨ 💕 ✌️ 💕 ┈💖━━━━━━━━💖┈ #NSFW #Nude #AI #SoloGirl #OpenLegs #Pussy #CuteAndNaughty #HotGirl #Sexy #Go… | ┈🌹━━━━━━━━🌹┈ ✨ [ #perfectgoddess] ✨ 💕 ✌️ 💕 ┈💖━━━━━━━━💖┈ #NSFW #Nude #AI #SoloGirl #OpenLegs #Pussy #CuteAndNaughty #HotGirl #Sex… | ↗Not briefed |
hn hn-ai | 10d | Show HN: Movement breaks that only show up while Claude Code is working | — | ↗Not briefed |
bluesky bluesky-ai | 10d | I'm basically here. I don't think we should be burning down indie studios over f**k-ups, more important is how they react going forward. If… | I'm basically here. I don't think we should be burning down indie studios over f**k-ups, more important is how they react going forwar… | ↗Not briefed |
bluesky bluesky-ai | 10d | yea its not great, also it's more of a me issue but i think the overuse of · is a LLM tell these days and generally bad design practice | yea its not great, also it's more of a me issue but i think the overuse of · is a LLM tell these days and generally bad design practice | ↗Not briefed |
bluesky bluesky-ai | 10d | I can’t understand: is this because your ISP user CGNAT and some other customer trained an LLM? | I can’t understand: is this because your ISP user CGNAT and some other customer trained an LLM? | ↗Not briefed |
bluesky bluesky-ai | 10d | I don't think I'm going to hash out a great discussion of this here, but although I don't think LLM assistance is actually a good long-term… | I don't think I'm going to hash out a great discussion of this here, but although I don't think LLM assistance is actually a good long… | ↗Not briefed |
rss techmeme | 10d | Sources and records: 11+ NSA and US Cyber Command personnel died by suicide over the 12 months to September, prompting White House and Congressional questions (Jake Bleiberg/Bloomberg) | Jake Bleiberg / Bloomberg: Sources and records: 11+ NSA and US Cyber Command personnel died by suicide over the 12 months to September… | ↗Not briefed |
arxiv arxiv-ai | 10d | Robust Task-Aware State Estimation from Pre-trained Perception | Agents that perform state estimation using pretrained perception models, e.g., Vision-Language Models (VLMs), face the fundamental cha… | ↗Not briefed |
hn hn-ai | 10d | Show HN: Tabichat chat demo, LLMs (with thinking) remain in their Pi harnesses | Demo for tabichat! Now with auto-renew for surviving long sessions.<p>By the way, anyone have any remote part time work available? Ne… | ↗Not briefed |
arxiv arxiv-ai | 10d | Mach2: Scientific Discovery Platform with Deterministic Execution Constraints | Language-model agents can formulate workflows and invoke specialized tools, yet compliance with natural-language operating rules is no… | ↗Not briefed |
arxiv arxiv-ai | 10d | SlipStream: Enabling HBF for GPU Systems by Mitigating the Latency and SRAM Overheads | Large language model (LLM) inference demands ever more memory capacity for larger models and batch sizes. High-Bandwidth Flash (HBF) r… | ↗Not briefed |
bluesky bluesky-ai | 10d | Claudes Georg, who works for Anthropic and has 20 quadrillion coworkers, is an outlier adn should not be counted | Claudes Georg, who works for Anthropic and has 20 quadrillion coworkers, is an outlier adn should not be counted | ↗Not briefed |
bluesky bluesky-ai | 10d | like to wave it away of just asking AI for some advice, nah cuz from their own statement "snippets from LLM responses were copied as a star… | like to wave it away of just asking AI for some advice, nah cuz from their own statement "snippets from LLM responses were copied as a… | ↗Not briefed |
hn hn-ai | 10d | The Segway PT was supposed to revolutionize travel. Disaster followed instead. | — | ↗Not briefed |
bluesky bluesky-ai | 10d | My thought right now is if the ney positive of something that used a.i. is greater than the negative, I can support it while still being hi… | My thought right now is if the ney positive of something that used a.i. is greater than the negative, I can support it while still bei… | ↗Not briefed |
bluesky bluesky-ai | 10d | ⚡ Tech giants like OpenAI, Google, and Anthropic are tightening sanctions, forcing Russians into a risky black market to access AI. This u… | ⚡ Tech giants like OpenAI, Google, and Anthropic are tightening sanctions, forcing Russians into a risky black market to access AI. T… | ↗Not briefed |
bluesky bluesky-ai | 10d | Trying out a new style for a commission I'm working on... #WIP #paleoart #mosasaur #tattoo #digital #art #noAI | Trying out a new style for a commission I'm working on... #WIP #paleoart #mosasaur #tattoo #digital #art #noAI | ↗Not briefed |