
What happened
At the Open Source AI for Science Summit, Meta's FAIR team said the Digital Brain Project, run with the Rothschild Hospital Foundation and University of Montreal, will raise its open brain-data target from 15,000 to 25,000 hours; it also released UMA Nitro.
Why it matters
More openly shared brain data and faster atom-level simulation tools could let university, hospital and industry researchers build on each other's work instead of starting from scratch, Meta says.
WHO IT HITSAcademic and hospital research teams in neuroscience, chemistry and materials science gain open models and data they can reuse, while high school chemistry teachers can now access UMA Playground through Discovery Education.
Summaries like this, every morning.
The summit reflects Meta's push to position openly shared models, tools and data as a common foundation for scientific work. The Digital Brain Project, run with the Rothschild Hospital Foundation and the University of Montreal, is the clearest example: by funding researchers to collect de-identified brain data and share it openly, Meta aims to give more teams the raw material to study human intelligence and neurological health.
Chemistry is the other focus. Meta released UMA Nitro, a faster version of its Universal Model for Atoms, and pointed to studies with Carnegie Mellon, UC Riverside and the University of Chicago. In one, UMA models identified many crystal forms of drug molecules without additional training data; in another, UMA simulated entire enzymes atom by atom. Those results suggest the models can handle tasks beyond their original training, though the body frames the enzyme work as potentially helpful rather than proven.
On the education side, UMA Playground is now available to high school chemistry teachers through Discovery Education, which reaches more than 4 million educators and 45 million students, while RealChem plans to integrate it into student activities. Meta also highlighted ATLAS, an open library formalizing 26 open-access textbooks and more than 45,000 computer-checked definitions and results, as a shared base for mathematics done in Lean.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
An independent developer published auroray, a roughly 200,000-parameter MLP filter that draws candidate object…

AgeAI, led by CEO Gen Mikami, joined the Gotanda Valley Acceleration Program 2026 from its September 17 kickof…

Jane Street published socratic-tutor, a 102-line text file in its agent_skills repo, on 2026年11月3日; a test on…

ConvergePartners said it agreed to adopt Paravellum's data-infrastructure technology to build a shared foundat…

Fixstars began offering GLM-5.3 on Fixstars Inference Cloud on November 20, using an OpenAI-compatible API; th…

Nvidia is in advanced talks to buy Inception, a two-year-old diffusion-based language-model startup, for more…
