THE APEX TIMES
At its Menlo Park summit, Meta’s FAIR team showcased open-source work ranging from an expanded “Digital Brain” data program to new models for materials and chemistry education.
Meta used its Open Source AI for Science Summit in Menlo Park to argue that freely shared models, tools, and data can speed up research across disciplines. The company’s Fundamental AI Research (FAIR) team said the two-day event brought together researchers from universities, hospitals, and industry to discuss how openness can help scientists tackle problems in healthcare, neuroscience, chemistry, materials science, and mathematics.
A centerpiece announcement was an expansion of Meta’s Digital Brain Project, which Meta described as a program to fund researchers collecting de-identified brain data and sharing it openly. Meta said it is increasing the project’s data target from 15,000 hours of brain data to 25,000 hours, positioning the effort as a way to support research into human intelligence and potential approaches to improving neurological health. The expansion was described as tied to partnerships that include the Rothschild Hospital Foundation and the University of Montreal.
Meta also highlighted work aimed at using AI to increase independence for people with disabilities. At the summit, the University of Pittsburgh presented an assistive-robotics system built on DINOv3 and SAM 3, two model families used for computer vision and segmentation. Meta said the team plans to demonstrate a robot that can help a person eat and is experimenting with allowing users to give instructions in their own words rather than relying only on tightly constrained commands.
In materials and drug discovery, Meta said it released UMA Nitro, a faster version of its Universal Model for Atoms. Meta described the Universal Model for Atoms approach as a way to simulate and reason about atomic and molecular structures. The summit also included examples of how UMA models can identify drug crystal forms: Meta said a collaboration with Carnegie Mellon and UC Riverside found many crystal forms of drug molecules without additional training data. Crystal forms matter because they influence how a pill dissolves and behaves in the body.
Meta further said the same UMA technology can be used for enzyme simulations. With the University of Chicago, Meta said it demonstrated the ability to simulate entire enzymes atom by atom. The company linked that capability to potential downstream uses in medicine, sustainability, and energy, framing the work as a way to help researchers understand how enzymes work and engineer improvements.
The summit extended beyond research labs to education tools. Meta said UMA Playground is designed to let users visualize chemistry more interactively, including building molecules, watching chemical bonds break and form, and exploring chemical reactions. Meta also cited distribution plans for classroom use: Discovery Education said the tool is now available to high school chemistry teachers with classroom guides, and Meta said Arizona State University and Carnegie Mellon plan to integrate it into student activities through a course called RealChem.
Meta quantified parts of that education footprint, saying Discovery Education reaches more than 4 million educators and 45 million students, while RealChem is used by nearly 100 institutions. Meta’s broader point was that the same AI-driven simulation and visualization approaches that help researchers can also help teach chemistry concepts in ways that are harder to replicate with static diagrams.
On the mathematical side, Meta described using open-source tools to verify proofs more efficiently. The company pointed to Lean, an open-source language for writing mathematics that allows computers to check each logical step. Meta said that could make it easier for mathematicians to build on one another’s work without having to re-check the same proof, including proofs contributed by colleagues, students, or AI assistants.
Meta also discussed an open foundation for mathematics collaboration. At the summit, FAIR researchers shared open-source projects, including ATLAS, which Meta described as an open library that formalizes 26 open-access textbooks and more than 45,000 computer-checked definitions and results. Meta said mathematicians guide the process, choosing research questions and ensuring the statements written in Lean reflect the meaning of those questions.
Meta’s announcement set out a clear thesis: openness can widen participation and reduce duplication in science workflows, from data collection to simulation to proof verification. Still, many of the summit’s claims were framed as demonstrations and early results rather than long-term performance metrics, commercial adoption, or quantified impacts on outcomes. The company did not, in the summit recap, provide independent verification, timelines for broader release schedules, or benchmarks showing how UMA Nitro, UMA Playground, or the robotics approach compares against alternative tools under standardized conditions.
A market report said the U.S. Securities and Exchange Commission has opened an investigation involving Nvidia as the chipmaker absorbs fallout from a sharp selloff. Investors now have to parse what the agency is looking at and how much the matter could weigh on disclosures and governance.
A newly announced partnership between Microsoft and Danaher aims to pair AI-assisted capabilities with cloud-based scientific instruments and lab software, including early tools that are already moving toward commercial use.
The investor said early demand outlines around Anthropic’s market debut suggest the broader technology sector still has investors’ attention, even after a difficult stretch for Nvidia.
The company alleges Apple’s “tap-to-share” behavior infringes its technology, asking a court to halt Apple iPhone and Apple Watch imports into India.
In a market selloff, a prominent analyst said investors may be reading a short-term pause too pessimistically and missing ongoing outlines from Nvidia’s latest results.
A market report says Apple is working on an AI assistant aimed at competing with Meta’s Muse, a move that could shape how the iPhone maker positions its AI features. Apple has not publicly confirmed the effort.
A new market note points to Amazon Web Services’ annual customer and partner conference as a familiar catalyst, asking whether the stock tends to rise in the month after re:Invent based on its last 10 events.
A report says Coatue Management built a stake of about 8 million shares of Salesforce during the third quarter, even as the stock has faced pressure. The move underscores how some investors are pairing patience with a large-scale bet on Salesforce’s performance.
A fresh investigation mentioned in market coverage helped trigger a wave of selling in NVIDIA’s stock, underscoring how quickly non-company catalysts can hit sentiment for mega-cap AI leaders.
A Yahoo Finance report says billionaire investor George Soros increased his Meta Platforms stake by 596%, even as the stock fell soon after. The episode highlights how hedge-fund-style disclosures can collide with day-to-day market sentiment.
Comments
Community guidelines. Comments may be removed for violations.
Loading comments...