Orchard is an open-source framework for the research community to train and evaluate AI agents across task types. It reduces complexity while supporting strong performance from smaller models by enabling researchers to reuse the same infrastructure. www.microsoft.com/en-us/resear...

LLMs do not get smarter just by remembering more. EvoLib turns experience into evolving knowledge, taking reusable skills and insights that help models learn and adapt across tasks long after deployment. www.microsoft.com/en-us/resear...

Researchers Jenna Butler, Jake Hofman, and Rebecca Janssen spend a lot of time studying AI’s influence on how people do their jobs. On the Microsoft Research Podcast, they share their thoughts on what the future of work looks like in the age of AI. www.microsoft.com/en-us/resear...

Today, we announce the second release of PazaBench, our benchmark for evaluating Automatic Speech Recognition (ASR) models across African languages.

Aurora 1.5 scales open forecasting; Flint redefines visualization; FlowDAgger accelerates adaptation. Plus: AI agents crack Rowhammer defenses and memory reshapes conversational AI. www.linkedin.com/pulse/22-new...

22 new weather variables, rapid post‑disaster building damage assessment, and robots that learn from human corrections.

In this issue: Aurora 1.5 extends open foundation models for weather and Earth‑system forecasting with 22 new variables, hourly resolution, and ensemble prediction Flint introduces a visualization int...

linkedin.com

Recalling a new password after hours of distractions can be hard. Not for a transformer. Microsoft CVP Doug Burger & AI researchers Subutai A. & Nicolo Fusi discuss the fundamental differences between human memory and machine intelligence. www.microsoft.com/en-us/resear...

Cryptographic code supports vital protections in modern computing systems. Learn how a new method helps verify code as developers write it while preserving speed and adaptability as it gets implemented and evolves. www.microsoft.com/en-us/resear...

Aurora 1.5 adds 22 more variables, hourly temporal resolution, and probabilistic ensemble forecasting to the Aurora foundation model, making it more useful for real-world weather, climate, and energy applications. www.microsoft.com/en-us/resear...

Short chart specifications are easy to write, but often produce uninspiring results. Flint is an open-source visualization language that offers a middle path, letting AI agents create expressive charts from compact, human-editable specifications. www.microsoft.com/en-us/resear...

Among Microsoft’s 100+ accepted papers, 3 oral presentations, and 1 expo demo at ICML in Seoul, highlights include Fara 1.5 computer-use agents, critique-resilient benchmarking, expanded protein ML benchmarks with FLIP2, and improved LLM reasoning stability. msft.it/6019vaCHU

AI agents often fail because their instructions, or skills, are manually modified with no guarantee of improvement. Learn how SkillOpt turns skill editing into a training process, making agent behavior more reliable without changing model weights: msft.it/6012vsvEs

SkillOpt turns AI agent skills into trainable assets

AI agents often fail because their instructions, or skills, are manually modified with no guarantee of improvement. Learn how SkillOpt turns skill editing into a training process, making agent behavio...

msft.it

AI agents can't remember past conversations. They must constantly reload or retrieve context, which grows less efficient as tasks get longer and more complex. Memora solves this with a scalable memory system separating what’s stored from how it's retrieved: msft.it/6018vs3gC

Researchers introduce generative causal testing, which translates black box models into clear hypotheses and verifies them in the scanner, revealing what specific brain regions respond to in language. msft.it/6011vUtRd

Talos was built to help resolve a major bottleneck in genomic medicine: human review time. The open-source system recovered 90% of in-scope diagnoses while surfacing just 1.3 candidate variants per patient for expert review. msft.it/6014vUBwK

On the Microsoft Research Podcast, Microsoft’s Jaime Teevan, Jenna Butler, Jake Hofman, and Rebecca Janssen explore how looking at AI as more than a tool for automating tasks people are already capable of might open the door to new possibilities. msft.it/6011vo8cX

Can you remember a new password after listening to five hours of stories? A transformer can. AI researchers Subutai Ahmad and Nicolò Fusi explore the fundamental differences between human memory and machine intelligence with Microsoft CVP Doug Burger. msft.it/6016vllRa

ResNet has received the Longuet-Higgins Prize at CVPR 2026, recognizing research with proven, lasting impact. A decade after its publication, residual connections remain foundational to how modern AI systems are built, with over 320,000 citations and growing. msft.it/6017vl6bl

30x faster analytics, GPU kernels generated automatically from SQL, AI matched to lab-grown tumor models for cancer treatment, and LLMs that learn across tasks without retraining. Dive into the latest issue of Research Focus: msft.it/6010vcYZ4

Project Ire examined a timely malware sample and determined its intent through reverse engineering—identifying LOTUSLITE characteristics even as most major EDR tools did not detect it. msft.it/6011viy4N

A new research effort called Encrypted Spaces explores an architecture for collaborative applications where data is encrypted, and operations are cryptographically verifiable. Learn more: www.microsoft.com/en-us/resear...

Encrypted Spaces - Microsoft Research

Encrypted Spaces is a research project (opens in new tab) exploring an architecture for building collaborative applications where data is encrypted and all operations are cryptographically verifiable....

microsoft.com

New research in Nature Methods from Project Ex Vivo shows AI models learn more from diverse cell states than from scaled datasets alone, a finding that could reshape how therapies are matched to patients. msft.it/6013vgE8l

A BTS look a first timer's experience at the MSR lab at Microsoft Build 2026—featuring the demos, the builders, and the conversations that made it worth the trip.

That’s a wrap on Microsoft Build 2026. Huge thank you to the researchers and team who brought it to life on the floor this week. If you missed what we shared, check it out here: msft.it/6015vjfld

What do the people who build the foundations for others to build on like to build themselves? The answers may surprise you, everything from sandwiches to the future. #MSBuild

You don’t have to be a developer. You can just build stuff. Scott Hanselman on his first walk through the Microsoft Research Lab at #MSBuild.

The sights and sounds of Microsoft Research at #MSBuild. Come see us at the Lab inside the Gateway Pavilion for Day 2.

A three‑month pilot in a Midwestern bottling plant shows what happens when AI moves beyond chat and into decision-making, where constraints shift, stakes are real, and answers must hold. msft.it/6015vjYUN

Day 1 of Microsoft Build is a wrap. A lot happened today, and we brought some of our favorite tools to the floor for developers to get hands-on with. Explore it all: www.microsoft.com/en-us/resear...

Weather forecasts thousands of times faster than traditional supercomputers. Hear from Kenji Takeda on Aurora at the Microsoft Research Lab at #MSBuild. Learn more: msft.it/6018vjGUA

Agentic experiences powered by small models that fit on your own device. Hear from Maya Murad on MagenticLite at the Microsoft Research Lab at #MSBuild.