Microsoft Research

VeriTrail, new from Microsoft Research, can detect AI-generated content that is not supported by the source text, trace the provenance of content from final output back to the source, and locate where errors were likely introduced. Learn more: msft.it/6012sMc3m

Stylized digital illustration of a multi-layered circuit board. A glowing blue microchip sits at the top center, with intricate circuitry radiating outward. Beneath it, four stacked layers transition in color from blue to orange, each featuring circuit-like patterns. Smaller rectangular and circular components are connected around the layers, all set against a dark background with scattered geometric shapes.

Designed to classify software without context, Project Ire replicates the gold standard in malware analysis through reverse engineering. It streamlines a complex, expert-driven process, making large-scale malware detection faster & more consistent: msft.it/6015s1tTN

Illustrated headshots of Peter Lee, Dr. Morgan Cheatham & Daniel Chen. The AI Revolution in Medicine, Revisited runs along the bottom of the graphic.

Medical education needs to evolve to meet the AI moment. In Ep. 9 of "The AI Revolution in Medicine, Revisited," Dr. Morgan Cheatham & Daniel Chen explore how clinicians in training use AI while navigating questions of responsibility & the future of care. msft.it/6010sJY94

In Episode 9 of "The AI Revolution in Medicine, Revisited," resident physician and venture capitalist Morgan Cheatham and med student Daniel Chen explore generative AI's impact on medical training, from classroom learning to clinical practice. msft.it/6011s1Qrd

Introducing Xinxing Xu. Xinxing Xu bridges AI research and real-world impact at Microsoft Research Asia - Singapore. Xinxing’s headshot sits to the right of the text.

Xinxing Xu is helping shape the work of Microsoft Research Asia – Singapore by turning advanced AI research into real-world solutions. Learn how he collaborates across sectors and disciplines to drive responsible innovation throughout Southeast Asia: msft.it/6016sEN2O

Microsoft Research Asia Launches Singapore Lab to Drive AI Innovation, Industrial Transformation, and Talent Development - Source Asia

So pleased to launch @MSFTResearch Asia – Singapore today. This takes our longstanding collaborations to the next level and gives new opportunities to advance AI to benefit humanity. We're grateful for the engagement of so many great people and orgs here. news.microsoft.com/source/asia/...

The image features three white icons on a gradient background transitioning from blue on the left to green on the right. The first icon, located on the left, depicts a network or molecular structure with interconnected circles. The middle icon shows a laptop with an attached mouse. The third icon, on the right, represents a hierarchical organizational chart with connected squares.

Semantic Telemetry helps LLMs run efficiently, reliably, and in near real-time. Learn about the engineering behind that system, including the trade-offs and lessons learned along the way—from batching strategies to token optimization and orchestration. msft.it/6011sEfI7

Crick researchers to collaborate with Microsoft Research AI for Science to map protein motion A new collaboration between researchers in the Crick’s Visual Biochemistry Lab and Microsoft Research AI for Science will combine cutting-edge laboratory experiments with artificial intelligence, to b...

Researchers in the Crick’s Visual Biochemistry Lab are teaming up with Microsoft Research AI for Science to combine cutting-edge lab experiments with AI to better understand and predict how proteins move and interact.
@franknoe.bsky.social

www.crick.ac.uk/news/2025-07...

Illustrated headshots of Amanda Craig Deckard & Kathleen Sullivan follows the quote. AI Testing and Evaluation: Learnings from Science and Industry runs along the bottom of the graphic.

In the “AI Testing & Evaluation” finale, Amanda Craig Deckard explores what Microsoft has learned about testing as a governance tool; the roles of rigor, standardization, & interpretability in testing; & what’s next for Microsoft’s AI governance work. msft.it/6017sDY5Z

Can AI help people stay healthier as they age by helping to predict disease? @peteratmsr.bsky.social and Dr. Eric Topol explore this and other topics in Episode 8 of “The AI Revolution in Medicine, Revisited.” msft.it/6019s6arx

CollabLLM teaches LLMs to collaborate with users Recipient of an ICML 2025 Outstanding Paper Award, CollabLLM improves how LLMs collaborate with users, including knowing when to ask questions and how to adapt tone and communication style to differen...

This approach helps move AI toward more user-centric and trustworthy systems. msft.it/6014Shf2i

Image llustrates the overall training procedure of CollabLLM. For a given conversational input, the LLM and a user simulator are used to sample conversation continuations. The sampled conversations are then scored using a reward model that utilizes various multiturn-aware rewards, which are then in turn used to update parameters of the LLM.

Recipient of an ICML 2025 Outstanding Paper Award, CollabLLM improves how LLMs collaborate with users, including knowing when to ask questions and how to adapt tone and communication style to different situations.

Illustrated headshots of Ciaran Martin, Kathleen Sullivan, and Tori Westerhoff. AI Testing and Evaluation: Learnings from Science and Industry runs along the bottom of the graphic

In “AI Testing and Evaluation,” former UK cybersecurity chief Ciaran Martin explores differentiated standards and public-private partnerships in cybersecurity, and Microsoft’s Tori Westerhoff examines the insights through an AI red-teaming lens.
msft.it/6010Shb5Q

Daphne Koller, Noubar Afeyan, & Dr. Eric Topol, leaders in AI-driven medicine, explore how AI is reshaping how we diagnose and treat disease—from early-stage drug discovery to clinical care—in Episode 8 of “The AI Revolution in Medicine, Revisited.” msft.it/6019S7p9t

Scalable emulation of protein equilibrium ensembles with BioEmu - Microsoft Research Following the sequence and structure revolutions, predicting functionally relevant protein structure changes at scale remains an outstanding challenge. Microsoft Research AI for Science introduces Bio...

Watch the full BioEmu explainer here: www.microsoft.com/en-us/resear...

AI is accelerating protein discovery Meet BioEmu-1 from Microsoft Research. This deep learning model can generate thousands of protein structures per hour, unlocking new possibilities for protein scientists and drug discovery and researc...

Learn more about BioEmu: www.microsoft.com/en-us/resear...

Also available: MD simulations generated to train BioEmu – more than 100 milliseconds worth of data of 1000s of protein systems and 10,000s of mutants. This dataset stands out for its combined protein sequence diversity and simulation length. www.microsoft.com/en-us/resear...

This also holds up for predicting stability changes of single and double mutants. These results indicate that the encoding of protein mutants still resolves enough differences to be predictive when fine-tuned with the right data.

BioEmu 1.1 improves ability to match experimental protein stability measurements with sampled protein structure ensembles with prediction errors below 1 kcal/mol, correlations >0.6 for a large protein stability test set, and train-test sequence similarities ~ 50%.

BioEmu 1.1 can emulate equilibrium distributions of millisecond-timescale MD at many orders of magnitude speedup, bringing GPU-years down to GPU-hours.

BioEmu 1.1 predicts functionally relevant conformational changes, including large-scale domain motions and local unfolding events + an increased success rate in predicting the formation of “cryptic” binding pockets.

BioEmu 1.1 is trained longer and more carefully in 3 distinct stages on vast data of protein structures, >200 milliseconds of molecular dynamics simulations, and 500,000 protein stability measurements.

BioEmu aims to emulate the ensemble of structures that a protein will adopt in an experiment or the cell. The ability of a protein to dynamically switch between distinct structures is a basis for its function.

Today in the journal Science: BioEmu from Microsoft Research AI for Science. This generative deep learning method emulates protein equilibrium ensembles – key for understanding protein function at scale. www.science.org/doi/10.1126/...

OSDI 2025 - Microsoft Research Microsoft is a proud sponsor of the 19th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’25) (opens in new tab). It will take place on July 7–9, 2025, at the Sheraton Boston in ...

This recognition highlights the impact of their work in advancing reliable systems research. Dive into this paper and explore all of Microsoft’s accepted research at OSDI 2025 on our conference site: www.microsoft.com/en-us/resear...

A promotional image for OSDI 2025 featuring a purple background with the OSDI 25 logo on the left. On the right is a partial view of a research paper titled “PoWER Never Corrupts: Tool-Agnostic Verification of Crash Consistency and Corruption Detection,” authored by researchers from the University of Texas at Austin, Microsoft, and MIT CSAIL.

Congratulations to Hayley LeBlanc,  Jay Lorch,  Chris Hawblitzel,  Cheng Huang and co-authors of “Power Never Corrupts: Tool-Agnostic Verification of Crash Consistency and Corruption Detection” for winning the Distinguished Artifact Award at #OSDI2025!

Photo Credit: Mark Stone/University of Washington

The method leverages the carbon-capturing potential of intact biomass while minimizing the need for energy-intensive processing. And thanks to machine learning, the researchers completed the optimization process in just 28 days—five times faster than conventional approaches. msft.it/6049SAKeJ

A new study from Microsoft Research and the @uwnews.uw.edu explores just that—introducing unprocessed seaweed in cement as a way to reduce its carbon emissions.

Cement is everywhere. It’s the second most-used material on Earth after water—and one of the biggest contributors to greenhouse gas emissions.

But what if we could use a natural green material to help build a more sustainable future?