Microsoft has unveiled a new artificial intelligence coding model that can run directly on personal computers, alongside security technology designed to prevent AI agents from accessing data without authorisation.
The company made the announcements at an event in San Francisco on Wednesday as it seeks to expand the use of AI on Windows computers and compete with Apple’s efforts to bring more AI capabilities to personal devices.
Microsoft also introduced the Surface Laptop Ultra, a high-performance computer powered by Nvidia’s RTX Spark chips.
The move is part of Microsoft’s strategy to turn Windows into a platform where AI agents can perform tasks such as writing computer code and handling complex business projects directly on desktops and laptops.
The company said moving some AI workloads from its Azure cloud data centres to powerful Windows machines could reduce computing costs while allowing businesses and consumers to use AI locally.
Microsoft’s new security system, called Microsoft Execution Containers, is designed to restrict AI agents from accessing data or performing unauthorised tasks on computers.
Pavan Davuluri, Microsoft’s executive vice president for Windows and devices, said the technology would allow IT departments to establish rules for AI agents, which the Windows operating system would then enforce on individual devices.
Microsoft Chief Executive Officer Satya Nadella said the company wanted to make desktop computers a secure environment for AI agents.
Nvidia CEO Jensen Huang, who appeared alongside Nadella at the event, said the technology could significantly change how AI agents are developed and deployed.
Microsoft also unveiled AI models that can previously only run through cloud services but can now operate locally on high-powered Windows desktops and laptops.
Davuluri said a version of Chinese AI company DeepSeek’s V4 model could run on computers with at least 60 gigabytes of memory and outperform OpenAI’s GPT-5 on some coding and reasoning tasks.
Microsoft said its Copilot assistant would continue to use cloud computing for more demanding tasks, while delegating some workloads to local models when privacy or cost considerations make on-device processing preferable.
READ ALSO:Â Google announces new AI subscription plan for advanced users
The company also said Meta’s Muse personal assistant would be available as a native Windows application and would use some of Microsoft’s new security tools.
However, the high cost of the hardware could limit adoption of locally operated AI.
The Surface Laptop Ultra will start at $2,599 and cost as much as $5,899 for a model with a 20-core processor, 128 gigabytes of memory and one terabyte of storage.
Analyst Anshel Sag of Moor Insights & Strategy said rising hardware costs could make local AI computing accessible mainly to consumers and businesses with larger budgets.
