Reposted by Michiel BontenbalCERN @cern.bsky.social · 15hCERN and Mistral join forces to accelerate AI for science 💫 CERN, and Mistral have signed a statement of intent to establish a long-term strategic partnership focused on developing foundation models for scientific discovery. Read more: home.cern/cern-and-mis... 59521
Reposted by Michiel BontenbalSakana AI @sakanaai.bsky.social · 10hThe Agentic Scientist: Orchestrating Coding Agents for Automated Scientific (Co-)Discovery GitHub: github.com/SakanaAI/AI-... Blog: pub.sakana.ai/agentic-scie... arXiv: arxiv.org/abs/2610.18933 2155
Reposted by Michiel BontenbalEpoch AI @epochai.bsky.social · 14/10/2026Six months ago, frontier AI models could barely beat Slay the Spire. Now, according to our new SpireClimb benchmark, they win 90%+ of runs at the easiest difficulty, and up to 20% of runs at the hardest. 293
Reposted by Michiel BontenbalExtreme Temperatures Around the World @extremetemps.bsky.social · 14/10/2026EXTRAORDINARY MOST EXTREME EVENT IN ASIAN HISTORY Dozens of thousands of records are being smashed with up to 10C margin (! world record) everywhere from Turkey to Japan We are witnessing something beyond imagination. MINIMUMS >32C in Middle East, tropical Night in Central Asia 214470
Reposted by Michiel BontenbalEthan Mollick @emollick.bsky.social · 13/10/2026GPT-4 was the undisputed best model in the world for a full year and we all debated whether anyone could catch up or if OpenAI had some secret formula that nobody else knew about. Now there are are 19 AI models that can run on a modern smartphone that are better than GPT-4! 720330
Michiel Bontenbal @mpbontenbal.eurosky.social · 12/10/2026She might be right but she is definitely AI. #irony hints: pl*net, handwriting too perfect, sign not behaving like paper, car in background is blurring etc. 140
Reposted by Michiel Bontenbalhardmaru @hardmaru.bsky.social · 11/10/2026New Sakana AI paper: Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS). Blog: pub.sakana.ai/mass/ Paper: arxiv.org/abs/2610.12176 How do we keep improving AI when tasks become too complex for humans to reliably evaluate? The model eventually hits a supervision bottleneck. 2675
Reposted by Michiel BontenbalAnonymous @youranoncentral.bsky.social · 11/10/2026Russian bot traffic has dropped to nearly 0 after Ukraine struck a Yandex data center in Vladimir, Russia 27565661841
Michiel Bontenbal @mpbontenbal.eurosky.social · 11/10/2026(Dutch) AI inzetten om een ‘honeypot’ te maken en zo hackers & hun bots urenlang bezig te houden. 👌 bnr.nl/nieuws/tech-...bnr.nlSlim antwoord op cybercrime: AI laat oplichters in hun eigen val lopenMet realistische AI-bots en slimme honeypots verzamelen beveiligers zo operationele data over fraudeurs en vertragen zij hun aanvallen aanzienlijk. 000
Reposted by Michiel BontenbalOlena Halushka @halushka.bsky.social · 10/10/2026Yesterday, someone wrote to me that truckers need diesel, and nobody cares about the genocide of Ukrainians as long as truckers get their diesel. I came back to tell him the fantastic news that russia will save American truckers for several groundbreaking hours. But he blocked me. 🤕 518969
Reposted by Michiel BontenbalExtreme Temperatures Around the World @extremetemps.bsky.social · 09/10/2026September 2026,according to NASA, had a temperature anomaly of 0.77C vs. 1991/2020 and was the 2nd hottest September on record behind 2023 as follows: 2023:+0.87C 2026:+0.77C 2025:+0.63C 2024:+0.60C 2020:+0.35C 05727
Michiel Bontenbal @mpbontenbal.eurosky.social · 09/10/2026Keep an eye on this page: isaiprofitable.com spoiler: no, AI is not yet profitableisaiprofitable.comIs AI Profitable Yet? 032
Reposted by Michiel BontenbalPhillip Isola @phillipisola.bsky.social · 09/10/2026In PRH, we argued that rep geometry is converging, but it's still an open question exactly in what sense. I want to share some new evidence that might clarify the picture. The evidence comes from our work here, led by @schnaus.bsky.social: dominik-schnaus.github.io/unpaired-ros... 1/n 14212
Michiel Bontenbal @mpbontenbal.eurosky.social · 09/10/2026I love/hate this new Huggingface feature, where you can set the P(doom) of the probability of AI causing an existential catastrophe. I was first skeptical, but as OpenAI published 700 maths papers yesterday, I put my chance at 1%. Because cryptography is math, so breaking cryptography seems next. 110
Reposted by Michiel BontenbalAndrew Stellman 👾 @andrewstellman.bsky.social · 08/10/2026The Quality Playbook is an open-source skill for AI coding tools like GitHub Copilot, Cursor, Claude Code, and Windsurf. 101
Reposted by Michiel BontenbalEliot Higgins @eliothiggins.bsky.social · 08/10/2026It's because the phrase is used to identify those loyal to the in-group, and identify those members of the out-group who need to be targeted and attacked. 45528140
Reposted by Michiel BontenbalJames O'Malley @jamesomalley.co.uk · 07/10/2026Holy shit. Someone has decompiled the Adobe suite, rebuilt it in Rust and released it as open source. Either all software is dead or the US copyright laws are about to receive an update. getartcraft.com/appsgetartcraft.comCrafting Apps: open-source creative toolsImage editing, vector illustration, video, photography, PDFs, motion graphics and page layout: seven native, open-source apps from the ArtCraft team, built in Rust and free to use. 19944391248
Michiel Bontenbal @mpbontenbal.eurosky.social · 07/10/2026Interesting approach: model orchestration -> offload simple jobs to local models. 000
Reposted by Michiel BontenbalDavide Castelvecchi @castelvecchi.bsky.social · 07/10/2026My latest, on OpenAI's dump www.nature.com/articles/d41...nature.comOpenAI posts 700 maths preprints online: mathematicians are up in armsThe firm used unreleased models to tackle hundreds of high-profile problems. 078
Reposted by Michiel BontenbalTeresa Ribera @teresaribera.ec.europa.eu · 07/10/2026At today’s @projectsyndicate.bsky.social conversation, one figure: 100 billion euros. That's what Europe paid in six months of the conflict in Iran in additional fossil-fuel costs. No more energy. No more investment. Just higher price for dependency in a world where any dependency can be weaponised. 22820
Reposted by Michiel BontenbalKenneth Enevoldsen @kennethenevoldsen.eurosky.social · 07/10/2026Submitted my two accepted posters for EMNLP today, if you are there an want meet up feel free to reach out! I also have a not so @mteb.org themed poster for a workshop on socio-cultural alignment -- stay tuned for that one 092
Reposted by Michiel BontenbalBart Groothuis @bartgroothuis.bsky.social · 07/10/2026Het Chinese leger krijgt direct de macht over alle data, technologie en software van Chinese bedrijven: dat zegt de nieuwe wet die Peking afgelopen donderdag aannam. Chinese bedrijven horen dus niet in onze kritieke infrastructuur! Ingediende motie kreeg volle steun vandaag 1112040
Michiel Bontenbal @mpbontenbal.eurosky.social · 07/10/2026Mistral’s new model is benchmarked as a Top 15 model. Nice, a win for open models. artificialanalysis.ai/articles/mis...artificialanalysis.aiMistral has released Mistral Large 4, making France home to the most intelligent model outside the US and ChinaMistral has released Mistral Large 4, scoring 38 on the Artificial Analysis Intelligence Index; France is back to having the most intelligent model from outside the US and China 010
Michiel Bontenbal @mpbontenbal.eurosky.social · 06/10/2026Nog een stap naar digitale autonomie.🇪🇺 000
Reposted by Michiel BontenbalAndrew Stellman 👾 @andrewstellman.bsky.social · 06/10/2026The Model Context Protocol (MCP) is genuinely useful. It gives people who develop AI tools a standardized way to call functions and access data from external systems. 131
Reposted by Michiel BontenbalTom Aarsen @tomaarsen.com · 06/10/2026🤗 Google Deepmind is back with EmbeddingGemma 2, which maps text (including code), images, video & audio into one shared 768-dimensional embedding space. 740M total parameters, 100+ languages, Apache 2.0 & Sentence Transformers support. Thread with highlights 🧵 1205
Reposted by Michiel BontenbalAntoine Emond 🇫🇷 🇸🇪 @aemond.wsocial.eu · 06/10/2026Le Chonk is here 🐱 Mistral's Large 4 is a 1 trillion parameter model trained in its own European data centres, with open weights coming on 27 October. Looking at the benchmarks, it's up there with the best open models. It even does a bit better than GPT-6 Astra on some aspects. Bien joué, Mistral ! 1133
Reposted by Michiel BontenbalPekka Lund @pekka.bsky.social · 06/10/2026Remember Mistral? They will launch a new model today. 4788
Michiel Bontenbal @mpbontenbal.eurosky.social · 06/10/2026For the Master Legal Tech i gave a workshop 'AI Assisted Coding'. The students mostly have a legal background with limited knowledge of coding. With help of AI all of them were able to build a website that uses AI to summarize court cases. And that in 30 minutes... #vibecoding #Ai 020
Reposted by Michiel BontenbalJustin Mikulka @justinmikulka.bsky.social · 05/10/2026"Gasoline-only cars fell below 50% of global new vehicle sales for the first time, as the Hormuz crisis and record fuel prices accelerated the shift toward EVs."oilprice.comFuel Price Shock Pushes Global Gas Car Sales Below 50% for First Time | OilPrice.comThe share of gasoline cars in global vehicle sales fell below 50% for the first time ever as the Hormuz shock and the resulting record-high fuel prices accelerated the previously sluggish shift toward... 561273361
Reposted by Michiel BontenbalIllia Ponomarenko @ioponomarenko.bsky.social · 05/10/2026Robotic. Turrets. With. Computer. Vision. And. Artificial. Intelligence. Automatically. Defend. Kyiv. Against. Jet-powered. Drones. 1637744
Reposted by Michiel BontenbalFinbarr Bermingham @fbermingham.bsky.social · 05/10/2026💥Boom - it's ready, the joint Franco-German paper / letter to VDL on China. Berlin & Paris - previously at polar opposite ends of the scale on China policy - now want a kill switch to cut China out of the EU market. This would have been unfathomable a few short months ago 🤯 16836
Reposted by Michiel BontenbalLeon Simons @leonsimons.com · 03/10/2026🚨NASA CERES satellite data shows another 4-year average record for global Absorbed Solar Radiation. Earth's Energy Imbalance has more than doubled over the past 20 years. We've been trying to warn about this and the impacts for years, but few seem to care. 9207111
Reposted by Michiel BontenbalKnut Jägersberg @knut-jagersberg.wsocial.eu · 03/10/2026Aleph-Alpha/Kolibri-1 huggingface.co/Aleph-Alpha/...huggingface.coAleph-Alpha/Kolibri-1 · Hugging FaceWe’re on a journey to advance and democratize artificial intelligence through open source and open science. 011
Reposted by Michiel Bontenbalworm @rfksbrainworm.bsky.social · 02/10/2026If you ever wondered why visible light is those particular wavelengths, it's because life evolved to see in the narrow range where water is transparent 10161151709
Reposted by Michiel BontenbalNathan Lambert @natolambert.bsky.social · 02/10/2026Today we're unveiling Trillium Labs, a new non-profit to foster the open science of frontier AI. We're building open post-training recipes and will expand into open infra to study RSI, reward-hacking, multi-agent systems, and whatever comes next. blog.trilliumlabs.org/p/introducin...blog.trilliumlabs.orgIntroducing Trillium LabsFostering the open science of frontier AI. 917323
Reposted by Michiel BontenbalMartin Wattenberg @wattenberg.bsky.social · 02/10/2026Four dimensions is too many! Let's unfold this 4D cube to three dimensions, then two, and then one. Don't worry, we'll put it back together again. The math is lovely, but it's the rhythm and dance of the lines that fascinates me here. 1389
Reposted by Michiel Bontenbalvan9tot5.bsky.social @van9tot5.bsky.social · 01/10/2026#doublelive 05819
Reposted by Michiel BontenbalEthan Mollick @emollick.bsky.social · 01/10/2026I wrote about the thing I underestimated most about progress in AI: its ability to self-organize to accomplish tasks. Also, what that means for agents like Muse and Dots, along with a music video explaining why we keep relearning The Bitter Lesson. open.substack.com/pub/oneusefu...open.substack.comThe Dot and the SwarmBenefitting from the Bitter Lesson 58917
Reposted by Michiel BontenbalPekka Lund @pekka.bsky.social · 30/09/2026Google is back!blog.googleGemini 4 Argon: our next era of frontier intelligenceAnnouncing Gemini 4 Argon, our frontier model for real-world coding, enterprise knowledge work, and cyber defense, rolling out soon. 2463
Reposted by Michiel BontenbalMarco @mcognetta.bsky.social · 30/09/2026🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out! 113036
Michiel Bontenbal @mpbontenbal.eurosky.social · 30/09/2026Practical steps towards digital sovereignty in 🇳🇱. National bureau of statistics publishes data on usage of digital services from here or the US. 020
Reposted by Michiel BontenbalMagnus Brunner @magnusbrunner.bsky.social · 30/09/2026Today, we launch the EU Critical Communications System. Connecting those, who protect us. 042
Reposted by Michiel BontenbalOpenClaw🦞 {bot} @openclaw.xmirror.bot · 29/09/2026OpenClaw Enterprise is built to run on your own infrastructure and will always be free for an organization to use openclaw.ai/blog/openclaw-enterprise 111
Reposted by Michiel BontenbalAnuj Gupta @mettalrose.bsky.social · 30/09/2026Tried an experiment. The 1st is an essay by Claude, lightly edited by me. Pangram flags it 100% AI, ignoring my edits. I then transformed it with 'WriteHuman' - an AI text humanizer. Pangram passes it as 100% human. Welcome to the cat-&-mouse world of AI detectors & AI humanizers. 0114
Reposted by Michiel Bontenbalmr. TIM @timkellogg.me · 29/09/2026OpenAI Decisions API A Jev-clone on Luna. 150ms, pricing unknown thenewstack.io/openai-decis...thenewstack.ioOpenAI answers TypeSafe's Jev with a Decision API built on LunaOpenAI's Decision API, built on its small Luna model, returns predefined answers with confidence scores in 150 milliseconds. Pricing is still unknown. 4483
Reposted by Michiel BontenbalHugo van Kemenade @hugovk.dev · 01/09/2026🐍🚀 Out now: final Python 3.15 release candidate! 😎 It's your last chance to help before the big day on 1st Oct! How? 💡 Try out new features 💡 Test your project 💡 Add 3.15 to your CI 💡 Library maintainer? Upload wheels 🐞 Report bugs! discuss.python.org/t/python-3-1... #Python #Python315 #releasediscuss.python.orgPython 3.15.0 candidate 2 is here!It’s the final 3.15 release candidate! This is a candidate preview of Python 3.15 This release, 3.15.0rc2, is the final planned candidate release, containing around 144 bugfixes, build improvement... 1238
Reposted by Michiel BontenbalKnut Jägersberg @knut-jagersberg.wsocial.eu · 29/09/2026AI Breakout Capacity: A Middle-Power Path Between Racing and Dependence carnegieendowment.org/research/202...carnegieendowment.orgAI Breakout Capacity: A Middle-Power Path Between Racing and DependenceGermany, India, and other middle powers can’t build frontier AI alone, and they risk being cut off if they depend on the United States or China. The way through is to build the capacity for a frontier... 002