Pro-tip: use a token budget when using /goal in the Codex CLI This is an experimental feature we're exploring, you might have to enable the flag. Ask Codex to do it for you. https://t.co/S70FxiUuG8

@umbertoleon For readers with technical background: https://t.co/auSkGusjhF

150 Top Mathematicians Issue “Leiden Declaration” Warning Against AI Hype and Overstated Reasoning Claims According to AFP, more than 150 leading mathematicians from around the world have signed the "Leiden Declaration" warning governments and the public not to be fooled by the hype and media hype of artificial intelligence companies about the problem-solving capabilities of this technology. These scientists believe that technology giants exaggerate the reasoning power of their models in order to attract astronomical investments of several hundred billion dollars. While these systems may not only lead scientific judgments to the dirt road by providing seemingly correct but fabricated answers, but also cause irreparable damage by entering areas such as weapons and mass surveillance. The authors of the declaration have stressed that the future of scientific research must remain guided and judged by the free will of human reason, not based on market timing and the commercial interests of a few large laboratories.

The last major keyboard moment was in 2007, when Apple put the keyboard on glass. Nearly 20 years later, we're introducing: Acti @openacti1, the Agentic Keyboard. Not another AI keyboard that fixes grammar. Not another voice keyboard that types faster. An invisible agent in every text field. It doesn't predict your next word. It takes your next action. 👇

Microsoft just released a new GUI agent on Hugging Face Sico-Evolution jumps from 39.8% to 82.9% Task Success Rate Outperforming GPT-5.4, Claude Opus 4.6, and Claude Opus 4.7 All from a 4B parameter model https://t.co/UNSCLF8VPT

【採用情報】Product Designerのポジションをオープンしました🐟 https://t.co/v5dUlw3HHP Sakana Chat🐟・Marlin🐬・Fugu🐡をはじめとした、プロダクトの体験をリードするProduct Designerを募集しています。 このような役割を担っていただきます。 ・コンセプト設計からUI、プロトタイプ、 ブランディングまで一気通貫でリードする ・エンジニアやリサーチャーと協働したプロダクト方向性の意思決定する ・Sakana AIならではのプロダクト体験のビジョンを構想する グローバルに利用されるプロダクトづくりに挑戦したい方、ぜひご応募ください🚀

【採用情報】Product Designerのポジションをオープンしました🐟 https://t.co/v5dUlw3HHP Sakana Chat🐟・Marlin🐬・Fugu🐡をはじめとした、プロダクトの体験をリードするProduct Designerを募集しています。 このような役割を担っていただきます。 ・コンセプト設計からUI、プロトタイプ、 ブランディングまで一気通貫でリードする ・エンジニアやリサーチャーと協働したプロダクト方向性の意思決定する ・Sakana AIならではのプロダクト体験のビジョンを構想する グローバルに利用されるプロダクトづくりに挑戦したい方、ぜひご応募ください🚀

best models < 128B params on SWE-bench_pro... @Alibaba_Qwen 3.6 27b still crazy, closely followed by @ornith_ 35B https://t.co/9BmWE8WGw1

S3-clients + Hugging Face Buckets = 💥 You can now query and write directly to HF Storage Buckets via the S3-compatible API Just one secret. Done. 🚀 https://t.co/id0DaVJkSS

You can now use 100s of tools with @huggingface Buckets, thanks to the new S3 API! Usually just one or two lines to change. https://t.co/V6PPIoWlh7

A study from @Stanford showed that 71.3% of chatgpt queries could be accurately answered by a local model. I suspect a major part of enterprise AI workloads could be run locally too for free (compared to the massive costs of frontier API cost). Also, it reduces the risk of these workloads being taken away from you because you own the models instead of renting them - which sounds like a good idea these days haha. That's why we're introducing the ability for everyone to filter AI models on @huggingface based on your local hardware. For me, there are 800k+ public models that fit on my M5 24GB and that I can use easily thanks to llamacpp. Let's go local AI!

new logo wyt https://t.co/EnVS05SSzM

@julien_c Let me onboard captain and hoist the colors ! (audio on) https://t.co/snq8Lor9QO

@julien_c Let me onboard captain and hoist the colors ! (audio on) https://t.co/snq8Lor9QO

@Drachs1978 https://t.co/cgKBPNc90e

@NoLimitGains https://t.co/cgKBPNc90e

@ruanyf https://t.co/cgKBPNc90e

@suraj_sharma14 https://t.co/cgKBPNc90e

@ricwe123 https://t.co/cgKBPNc90e

@k1rallik https://t.co/cgKBPNc90e

@claudeai https://t.co/cgKBPNc90e

Claude: You are right. I did ignore your instructions. 💀 https://t.co/I2TEJhRkTl

Ford's experience is a good reminder that AI doesn't replace expertise overnight. Sometimes the best results come from combining AI with people who have years of experience and judgment. Technology is powerful. Human know-how still matters. https://t.co/8DYScUdlgW

https://t.co/3X1rbWvzCh

A living map of everything your Hermes agent has learned; every memory & skill Press play to watch it unfold, @NousResearch Copy yours into a code, paste a friend's to explore theirs https://t.co/UEYuYKNtKy

We've spent a lot of time talking about the computing power AI needs. Now another challenge is coming into focus: keeping that infrastructure running in a world of extreme weather. Building AI is one thing. Building it to be resilient is another. https://t.co/XSbj7YplEl

What if instead of autoresearch reflecting incremental progress from a single person, it reflected *hundreds* of researchers’ progress, updated live, with every run’s data point being interactive and reproducible? You could survey the full space of explored ideas, see which changes actually move the benchmark, and fold the most promising ideas into your next run. Labless (https://t.co/B00yoWPOK3) is my attempt at this. It can support thousands of simultaneous training runs across contributors around the world, each learning & improving together to hillclimb on a fixed benchmark, validated using a standardized codebase scaffold. Importantly, all valid training runs auto-submit to Labless, so failed experiments are as visible and reproducible as the successes. We have an agentic API that lets you or your coding agent study every run submitted so far so you don't waste time investigating something someone already tried, and you can source novel ideas from other contributors. Labless already hosts our Nanopath challenge to build the best pathology foundation model that trains in 1 hour on a single GPU. We are hoping to expand the platform further and support a suite of open-source research challenges. If this resonates with you, if you think the path to research innovation lies in open-source collaborative hill-climbing, please reach out.

One study found Musk’s closing of the U.S. Agency for International Development could cause the deaths of millions of children. https://t.co/qaQMoCPf82 (Photo: Associated Press) https://t.co/oOJ7N69g93

Recently we had an earthquake that mostly affected northern Venezuela. There are a lot of people helping right now. Within that help, a lot of tech enthusiasts & vibecoders rushed to create help portals, & that's good. The only issue is that information is scattered across several portals & it's hard to keep track of all of them. So, what I did was build a data ingestion pipeline that is self-directed with Gemma, with several sources of truth, as well as medical records that are publicly shared so their relatives can check & see that they're safe and alive. So, to get all the information in one place, I built https://t.co/ZqCn0nsfkd, which exposes an AI agent powered by Gemma 4, with the speed of @cerebras as part of my entry for the 24h Cerebras hackathon. It has the following characteristics: a live map with building damage based on a map by Microsoft + Google, a public repository for anyone interested, + an AI agent to search for relatives to update information autonomously in a loop, + information related to collection centers if you're interested in making donations.

ChatGPT/Claude are just amazing for travel planning. To the minutest detail of which side of the train to sit to get the best views. Which places would be great to go with a kid. What not to do. Picks the best offbeat locations & experiences. https://t.co/IQVrwHS6nj

here you go: https://t.co/qw4rSHBCcF

In the UK, Police officers would have special "cop nights" run by the Pakistani rape gangs where they were allowed to gang rape grooming gang victims as a "Thankyou" for turning the other way. Bring back capital punishment 😡 https://t.co/QObbm7pbol