Born out of frustration that his children preferred video games to basketball practice, Craig Moody developed Shoot 360 to give the game a new, high-tech boost that turns every shot and pass into valuable data. At Shoot 360’s facilities, players use interactive drills, instant feedback and insights drawn from more than 600 million shots – along with human coaching – to fine-tune their skills. Thanks to Microsoft technology, everything from shooting arc to ball handling is tracked, helping athletes identify areas for improvement while earning points and climbing global leaderboards.

A new Microsoft research model, CARE-X, could bring several radiology tasks under one roof, generating reports while also producing diagnostic assessments. A study found CARE-X performed well when researchers tested it on a range of chest X-ray tasks and on real-world patient data. In related experiments, researchers found the AI model performed better on some conditions when it could measure and calculate key values, instead of estimating them from the image.

Microsoft says its new MAI-Code-1.1-Flash model is delivering better code while running faster and costing far less than the version released in June. Now powering GitHub Copilot, the model showed gains on command-line coding tasks and .NET software development work — areas developers said mattered most. Microsoft also reported increases in code survival and return visits, since the model uses fewer tokens to complete tasks.

Microsoft is tracking DeadLock, an emerging ransomware operation that has claimed more than 80 victims across industries and regions, with more than half in Europe. Microsoft’s analysis shows DeadLock uses blockchain-backed services and a decentralized recovery system that can help parts of its operation remain available, even when disrupted.

Imagine an office worker clicking on a phishing link at 2:13 a.m. As the IT team sleeps, an AI agent detects unusual login behavior and disables the compromised account. Then, it checks to see if malware spread to any other devices or accounts, opens a ticket, sends an alert to the security team and drafts an incident summary.

This is the beauty of AI agents, and getting started is easier than it sounds. This guide walks through the key steps for creating a basic agent, so you can go from idea to working prototype without writing any code.

Two young Indonesians found new paths into the digital economy through Microsoft Elevate. Yusuf, a forestry researcher from Makassar, used AI and machine learning to strengthen geospatial analysis for flood-risk research. Lano, an informatics student from Tangerang Regency, turned to cloud technologies and GitHub Copilot to overcome device limitations and build real-world projects. Their experiences underscore how access to learning can expand opportunity, regardless of background or circumstance.

To help small and medium-sized businesses adopt AI, Microsoft is rolling out “Copilot in 30,” a 30-day trial of Microsoft 365 Copilot Business. It’s designed to give firms that have fewer than 300 employees a chance to see how AI can be incorporated into their business, with resources and guidance for how to use it. The initiative is part of broader investments in incentives, training and tools designed to move customers from evaluation to adoption and, eventually, more advanced AI workflows.

Microsoft’s newest datacenter region just went live in Hyderabad, India. The company now operates four cloud regions in India, giving it the country’s largest hyperscale cloud presence. For companies in the area, the move means lower latency and more capacity for AI use.

“Whether it is a Frontier Firm already running AI at scale or an enterprise taking its first steps, the question we hear most often is: how do we move faster from experimenting with AI to creating durable business value with AI?” Puneet Chandok, the president of Microsoft Indian and South Asia, said in a statement. “That requires trusted infrastructure close to where data lives, teams work and decisions are made.”

Microsoft is introducing Orchard, an open-source framework designed to make advanced AI agents easier to build and study. Instead of relying on proprietary infrastructure, Orchard provides a reusable environment for training and evaluating agents that can write code, navigate the web and handle productivity tasks. The results suggest that relatively small open-source models can tackle complex real-world work, helping lower the barriers to agentic AI research and development.

Leaders across industries are finding that AI’s biggest impact isn’t about doing work faster; it’s about helping people do more meaningful work. In conversations at Microsoft AI Tours around the world, leaders from organizations including The Salvation Army and Teach for America described a shift from AI experiments to everyday use, with employees using the technology to solve problems, learn new skills and bring ideas to life, while keeping human judgment and expertise at the center.

Artificial intelligence is already helping clinicians spot patterns in diagnostic images and detect signs of disease. But many of today’s pathology AI systems are built for a single task and must be rebuilt for each new application.

A study recently published in Nature Medicine  describes a different approach aimed at advancing research in this field. Researchers from Microsoft Research and Paige, now part of Tempus, developed PRISM2, a pathology foundation model trained on both tissue images and language based on real pathology reports. The goal was to help AI learn from the large amount of available text data and create a system that could support future pathology research and development.

A subsection of the Russian cyberespionage group Midnight Blizzard has compromised hotel, conference center and other guest Wi-Fi networks to redirect travelers to phishing sites and malware disguised as software updates. Microsoft, which has tracked Midnight Blizzard for years, says the CaptiveCrunch campaign combined AI‑supported operations, credential theft and device code phishing in an effort to gain access to accounts and collect data from Microsoft 365.

Finding software vulnerabilities is becoming increasingly important as AI makes cyberattacks faster and cheaper. Microsoft’s new MAI-Cyber-1-Flash model is designed to find complex software flaws and is built into the company’s MDASH vulnerability identification and remediation system. Microsoft says the combined system outperforms leading competitors on the CyberGym benchmark and cuts costs by 50%. It will help power Perception, a new agentic security platform.

At Singapore’s National Trades Union Congress, AI readiness begins with leading by example. Chief transformation and technology officer Siow Shong Seng uses Microsoft 365 Copilot and agents in his daily work, helping shape how staff adopt AI. He’s also showing the way for over 1.4 million union members navigating new ground in AI-forward Singapore.

Microsoft says Microsoft 365 Copilot has surpassed 30 million paid seats, with net seat additions more than doubling quarter over quarter, as organizations increasingly incorporate AI into everyday work. The company says AI is shifting from a productivity tool to a participant in workflows, including the adoption of agent-based products such as Copilot Cowork and Microsoft Scout. Microsoft says future progress will be measured not just by AI adoption but by how work itself changes.

Tokenmaxxing might be the AI story of the summer, but token efficiency is the next big focus across the industry. Microsoft’s latest approach is to get better results from AI tokens, which are the small units of text models process, rather than simply using the maximum amount of them. The company says a series of specialized models, tuned for specific tasks, are matching or exceeding the performance of larger alternatives while cutting costs, in some cases by 50% to 90%.

In Mexico, AI skilling is moving beyond theory and into people’s daily lives. Through Microsoft Elevate, these four young entrepreneurs and educators are applying AI skills to challenges they know firsthand: water scarcity in Ciudad Juárez, recruiting efforts and productivity in Chihuahua, recycling in Querétaro and personalized learning in classrooms. They are among the 3.5 million people reached by the initiative in Mexico. Their work shows how personal stories can shape practical innovation, and how access to AI training can help local builders turn curiosity, community needs and real-life experience into tools with real impact.

In a fiscal year defined by AI’s shift from promise to practice, businesses moved beyond experimentation to deploying AI for real-world business outcomes. They unlocked innovation and created new opportunities for growth, moving beyond efficiency gains to focus on human ambition and embed AI at the core of how they operate. Judson Althoff, CEO of Microsoft Commercial Business, looks back at Fiscal Year 2026 and the emergence of Frontier Firms. Read more on the Official Microsoft Blog.

The Premier League is rolling out a new tool to help football fans around the globe manage their fantasy teams. The Fantasy Premier League Companion powered by Copilot is built using Microsoft Foundry and official Premier League data to offer advice, answer questions and provide suggestions to managers as they build and maintain their squads. It’s all in time for the start of the season, which kicks off Aug. 21.

As cyberattacks become faster, cheaper and increasingly driven by AI, Microsoft says security needs to evolve beyond existing systems that were built for a human-scale threat landscape. Project Perception, a new agentic security system entering public preview Aug. 3, uses specialized teams of AI agents acting at machine speed to identify risks before attackers can exploit them, investigate and evaluate threats and then take corrective actions. Microsoft says the system will help organizations strengthen defenses while keeping humans in control.

Microsoft’s latest threat report found that the company’s efforts to disrupt the Tycoon2FA phishing platform in March continued to ripple through the second quarter, driving a sharp decline in related phishing activity. But even as some email threats fell, attackers increasingly turned to Microsoft Teams-based social engineering and voice phishing. The company detected 7.6 billion email phishing threats during the quarter, with credential theft remaining the primary goal.

XBOX is bringing some of its oldest games to new devices. Starting today, four original XBOX titles are playable on PC and supported handhelds for the first time, including the ROG XBOX Ally. Launching in early release, the effort begins with four classic games and new features such as graphics customization and cloud support. Microsoft says the program is part of a broader push to preserve older games and make them playable across more devices.

A year after launching an effort to build AI models in-house, Microsoft says its MAI models are now powering products including Bing, PowerPoint, OneDrive, Dynamics 365 and Azure. The company introduced two new variants — MAI-Image-2.5-Pro for high-fidelity image generation and MAI-Voice-2-Flash for faster voice applications — while expanding use of its own models across consumer, enterprise and healthcare products.

An animal shelter forecasting where foster homes will be needed before kitten season. Researchers studying a neurodegenerative disease that affects hundreds of thousands of people. A small team spending more time mentoring and less doing paperwork.

These stories don’t have much in common on the surface, but together, they point to a broader shift: AI is helping nonprofit organizations across the globe meet pressing challenges, make better use of limited resources and expand their reach. Read on to meet three nonprofits using Microsoft’s AI to help solve different problems.

The future of AI depends on more than bigger models. The real challenge is getting the best results at the right cost, says Microsoft. Its MAI models are trained inside products such as GitHub Copilot and Excel, learning from the tasks people actually perform. Early results show these smaller, specialized models can match or outperform larger, more expensive alternatives in common workflows while using fewer tokens and lowering the cost of deployment.

Microsoft and Databricks are expanding their decade-long partnership into the 2030s to help the companies’ customers use AI more effectively while maintaining governance, security and cost controls. Databricks will expand its use of Azure technologies, while Microsoft will further integrate Databricks’ data and AI tools into its products, helping connect AI tools to trusted business knowledge organizations rely on, from customer data to internal processes.

Manulife is expanding its longstanding partnership with Microsoft through a new five-year agreement aimed at helping the company use AI more broadly. The financial services provider will roll out Microsoft 365 Copilot to more than 30,000 employees and deploy Microsoft Agent 365 to help oversee AI agents across the business. The companies said the move will support innovation while strengthening AI governance and security across Manulife’s global operations.

Much of what AI knows about the world comes from data most people never see and from technical processes that determine what’s included or left out.

As AI becomes increasingly influential in daily life, a new platform aims to help communities participate in that foundation being built behind the scenes to ensure they’re portrayed accurately and fairly, especially in imagery. Called the Community Library Creator, the Microsoft research effort gives groups a structured way to help shape how they show up in AI-generated pictures.

Microsoft is deepening its commitment to the U.S. Department of Energy’s Genesis Mission, an effort to use AI to accelerate scientific discovery, with a $60 million investment and the launch of SPARK, a new hub for scientific collaboration. The effort brings together national laboratories, AI, advanced computing and emerging quantum technologies to support research in areas including energy storage, biosecurity, materials science and nuclear energy.

As Mistral expands its AI computing capacity in Europe, Microsoft is deepening its partnership with the French AI company through a multibillion-dollar infrastructure agreement. Mistral’s models will be available across Microsoft services, including Microsoft Foundry, Copilot Studio and Azure, giving customers more options for how and where they run AI. The move is designed to help organizations access advanced AI while maintaining control over data, operations and critical workloads.