🔬 Weekly Science Read 🧬 A virus, designed entirely by AI. @FT covers the first-ever feat, and the biosecurity gap it exposes: tinyurl.com/p9szhnbs

📺 Tune in now! Prof. @Yoshua_Bengio is addressing the @UN Security Council on frontier AI safety and the risks of losing control over advanced systems. webtv.un.org/en/asset/k1c/k1…

🔬🌍 How do we get science to the negotiating table early enough to shape decisions? Ahead of #UNGA81 High-Level Week, @UNUCPR & @FrontiersIn brought together policymakers, scientists & experts to explore how evidence can better inform multilateral action.

🔬 Weekly Science Read 🤖 Can you regulate what you don't understand? Board Member @UNUniversity Rector Tshilidzi Marwala on AI literacy in governance. Read it here: tinyurl.com/et74p22v

🔬 Weekly Science Read 🌪️ Predicting extreme weather needs AI that understands physics, not just data. @AnimaAnandkumar & @Caltech /@nvidia show how. tinyurl.com/5n79w46u

🔬 Weekly Science Read 🦠 Beating antimicrobial resistance needs more than new drugs, it needs One Health. Read the new editorial by Ramy K.Aziz & co. tinyurl.com/yrwnyy2k

🔬 Weekly Science Read 🏛️ Good science ≠ good policy. Science Day 2026 (ISC) says the gap isn't evidence, it's institutions & trust. Read the recap here: tinyurl.com/34m35nmt

Tackling a 60-year-old challenge in quantum chemistry: making density functional theory scale nearly linearly with system size. This has huge implications for opening the door to realistic systems that have traditionally been too expensive to simulate. AI has attempted to accelerate these calculations, but models generally struggle to extrapolate, particularly to systems larger than those seen during training. Unlike text or images, quantum-mechanical training data is extremely expensive to generate. We present a single unified AI model that performs quantum-mechanical simulations of both molecules and materials in quasi-linear time. Using a novel Fourier neural operator variant, we learn the underlying Kohn–Sham equation map to produce physics-informed, self-consistent answers. Unlike prior AI approaches that directly predict chemical properties, our model works through intermediate steps to improve difficult predictions, resembling inference-time reasoning in large language models. To demonstrate its scalability, we run a self-consistent calculation of a magnesium dislocation with about 80k electrons on a single GPU, which previously required about 7k GPUs. tensorlab.cms.caltech.edu/us… @DanishK42 @Caltech

Excited to see @Reuters cover the launch of our startup Accelerated Understanding. We are training large scale AI models that can simulate and understand physics to invent and discover. Our models understand the world directly in 4D (3D + time) and across physical phenomena. Going full 4D requires massive context length, we have pushed it to a Trillion in training and exceeding 5 Trillion at inference. AI giving you a bigger haystack of ideas doesn’t help. The bottleneck for new inventions and discoveries is shifting from ideas to the ability to test them. With AI that can simulate and understand physics we are directly attacking this bottleneck. People have been trying to do this for a while now, but usually by taking shortcuts. Narrow surrogates are great if you happen to have enough of precisely the right data and your design loop stays in distribution. Video models look fantastic but sweep physical accuracy under the rug, and some static world models cut out physics altogether. A lot of interesting physics isn’t visual. What does not cutting corners look like? Space stays 3D and you also have time: so 4D in total. You also need multiple physical modalities in the same model, not just things you can see. That’s what we’ve built. Scaling is the primary ingredient to make this work. To represent the world you need sufficient context, which in our case grows in 4 dimensions. Individual samples get so big they don’t fit into single accelerators or even full nodes anymore. We’ve developed architectural tricks to make it work. We’ve pushed our models to 1T parameters during large scale pre-training and are able to train at up to a Trillion context when needed and do inference exceeding 5 Trillion context without any sub-sampling or patching. Building on prior successes of AI weather forecasting, fusion simulation, design of medical devices, drugs and chips, we wanted to see if scale and universality can benefit AI for physical understanding. With our teams’ experience in large-scale infrastructure and model training we’ve been able to pull it off. reuters.com/business/ai-foun… acceleratedunderstanding.com… @accelerated_u @bjenik

Are we ready for the next pandemic? 🦠 🎙️ In Episode 4 of Science F⏩rward, @WHO Chief Scientist @SCBriand and @ProfHeidiLarson discuss what we learned from COVID-19, how science can help us prepare and why trust matters. linktr.ee/scientificadvisory…

🆕 🌟 Southern Voice's Global Policy unit is also becoming stronger, with a new addition to support our international work and reach! It is our pleasure to introduce you to… Meet Gourangi Kumar, our new Policy Officer! Gourangi brings her expertise in cross-country analytical policy work, covering areas such as education, gender equity, entrepreneurship, and regional cooperation frameworks. 💡 Her professional background covers policy research consultancy, specifically in institutional mapping of regional bodies, development financing, and international education partnerships 🎓 Education: Bachelor's degree in Political Science and Government, and Master's degree in Public Policy 📍 Location: India and France 🇮🇳🇫🇷 Gourangi's direct expertise in multilateral cooperation frameworks, Global South policy debates, and cross-regional stakeholder engagement will be a great addition to our team and work. Welcome, Gourangi! 🤗

🎙️ In the new Science F⏩rward, @UNEP’s @andersen_inger and @jrockstrom explain climate tipping points, the risks of crossing them, and why every tenth of a degree of warming we can prevent makes a difference. We also hear from @antonioguterres. 🎧: linktr.ee/scientificadvisory…

It was an absolute pleasure to meet in person @ScienceBoard_UN and @UN team, hosted by Fondazione Compagnia di San Paolo at the beautiful and historic Academia delle Scienze di Torino, co-located with the oldest Egyptian musuem, in Turin. I am proud to contribute to the upcoming briefs emphasizing how #AI is making an impact in diverse areas like climate modeling, agriculture and chemistry. This broad view of AI beyond language models is essential and the public needs to be made aware of the immense benefits AI is bringing about. Here's my short interview from Turin: piped.video/rJi3msuB-rw?si=sN8Z…

Opportunity: The International Science Council (ISC) is seeking applications from outstanding early- to mid-career researchers to serve as a Civic Science Fellow with a focus on science communication. Hosted virtually by the ISC, the 18-month Fellowship provides an opportunity to explore how multilateral organizations can communicate evidence more effectively in international policy processes and strengthen the role of science in multilateral decision-making. More info: buff.ly/NfNCs8h

🔬 Weekly Science Read 🌍 For 25 years, @WHO’s Research4Life initiative has expanded access to scientific knowledge, reaching 12,000+ institutions across 120+ countries. 📚 A look at the progress made — and what comes next. 🔗 tinyurl.com/24w4h9bk

J’ai discuté de l’importance d’un meilleur encadrement technique et sociétal de l’IA avec @Anouch du journal @LeTemps à Genève la semaine dernière dans le cadre des sommets de l’ONU et de l’UIT sur la gouvernance de l’IA. Entrevue complète : letemps.ch/cyber/yoshua-beng…

🎥 "AI and biotechnology hold enormous promise, but they must be developed responsibly." @Yoshua_Bengio opened the @ScienceBoard_UN's Turin panel on AI-biotechnology convergence with a special video message. @AnimaAnandkumar @SCBriand Abdoulaye Djimdé @tmarwala Thomas Südhof

The creation of a regulatory body to enforce and implement standards, as suggested here by @demishassabis or by @AnthropicAI last month, would be a huge step in the right direction to make AI development safer. Regarding the idea of an initial phase during which models are shared on a voluntary basis: I understand the argument from a pragmatic standpoint, but we absolutely need a clear and precise roadmap to transition from a voluntary to a mandatory model.

📚What counts as a scientific contribution? @ISC has launched a global survey to explore how science is recognized—and what valuable contributions may be overlooked, from policy advice and data sharing to education and community engagement. 🔗 tinyurl.com/3uusht9h

The @ISC is running a survey asking: "What forms of scientific contribution go unrecognised today?" This is your change to have your say on what non-orthodox science outputs are not being adequately valued by current reward and recognition systems? A great discussion to be having! buff.ly/nDkPn7n