# Microsoft AI

Microsoft AI (MAI) is an American artificial intelligence research laboratory and division of Microsoft, founded in 2024, overseeing consumer AI products like Microsoft Copilot and developing in-house models.

Microsoft AI (MAI) is an American artificial intelligence research laboratory that operates as a division of Microsoft. It oversees the company's consumer artificial intelligence products, including Microsoft Copilot, and is headquartered at Microsoft's campus in Redmond, Washington, with additional offices in New York City, London, Mountain View, Beijing, Suzhou, Hyderabad, and Zurich. The division was established in March 2024 to consolidate Microsoft's AI efforts under a single leadership team, bringing together research, product development, and engineering for consumer-facing AI services.

MAI builds on decades of Microsoft research in [artificial intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) and [machine learning](https://www.wikiprompt.org/wiki/machine-learning), including earlier conversational models such as DialoGPT. The division has since developed its own proprietary models, including voice, image, and code generation systems, while continuing to integrate [OpenAI](https://www.wikiprompt.org/wiki/openai) models into products like Copilot. Its leadership includes Mustafa Suleyman, a co-founder of [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind) and [Inflection AI](https://www.wikiprompt.org/wiki/inflection-ai), who serves as Executive Vice President and CEO, and Karén Simonyan, also an Inflection AI co-founder, as Chief Scientist.

## Founding and Early History

Before MAI's creation, Microsoft researchers had already released several notable conversational AI models. DialoGPT, introduced on November 1, 2019, was a dialogue generation model based on the GPT-2 architecture, trained on 147 million Reddit conversations. This work laid the groundwork for later products and research.

MAI was founded on March 19, 2024, as a dedicated division to oversee Microsoft Copilot and other services previously developed by separate teams, including the Edge browser and Bing search engine. The division also provided resources and co-developed features for the Microsoft 365 application suite. Microsoft's previous AI team was absorbed into the new organization, with its employees joining MAI. Mustafa Suleyman was appointed CEO, and Karén Simonyan became Chief Scientist, bringing experience from their prior roles at Inflection AI and DeepMind.

## Expansion and Product Integration

On April 7, 2024, MAI opened an AI hub in London, led by Jordan Hoffmann, who had previously worked at DeepMind. Staff who had worked on AI at Microsoft before MAI's founding were relocated to the new hub, signaling the division's international research ambitions.

In May 2025, following the shutdown of Skype, the messaging service GroupMe was moved under MAI and integrated with Copilot, expanding the division's consumer product portfolio. This move reflected Microsoft's strategy to embed AI across its communication tools.

On June 30, 2025, MAI announced development of a tool called "Microsoft AI Diagnostic Orchestrator," which the company claimed was four times more successful than human doctors at diagnosing complex ailments. The tool was planned for potential release across MAI's products in the future, highlighting the division's focus on healthcare applications.

## In-House Model Development

Historically, Microsoft's AI products, including Copilot, relied heavily on models developed by OpenAI. However, on August 28, 2025, MAI announced its first in-house models: MAI-Voice-1, a speech model, and MAI-1-preview, a large language model. This marked a strategic shift toward reducing dependence on external providers and building proprietary capabilities.

On October 13, 2025, MAI introduced MAI-Image-1, a text-to-image generation model, further diversifying its model portfolio. In June 2026, the division released seven additional in-house models, including MAI-Image-2.5 and MAI-Code-1-Flash, along with five others, demonstrating rapid progress in [generative AI](https://www.wikiprompt.org/wiki/generative-ai) across modalities.

These models are built on advances in [deep learning](https://www.wikiprompt.org/wiki/deep-learning), [neural networks](https://www.wikiprompt.org/wiki/neural-network), and [large language models](https://www.wikiprompt.org/wiki/large-language-model), leveraging techniques such as [transformers](https://www.wikiprompt.org/wiki/transformer), [multi-head attention](https://www.wikiprompt.org/wiki/multi-head-attention), and [positional encoding](https://www.wikiprompt.org/wiki/positional-encoding). The division also employs methods like [RLHF](https://www.wikiprompt.org/wiki/rlaif) and [model pruning](https://www.wikiprompt.org/wiki/model-pruning) to optimize performance and efficiency.

## Superintelligence Division

On November 6, 2025, MAI announced the creation of a superintelligence division and team, led by Mustafa Suleyman. The vision was described as "humanist superintelligence," with three primary applications: serving as an AI companion to help people learn, act, be productive, and feel supported; offering assistance in the healthcare industry; and creating "new scientific breakthroughs" in clean energy. This initiative positioned MAI at the forefront of advanced AI research, aiming to push beyond current capabilities.

## Copilot Division and Reorganization

In March 2026, MAI announced the creation of a Copilot division, which would include Copilot, Microsoft 365, and consumer apps such as Edge and Bing. Jacob Andreou was appointed to head the division, reporting directly to Microsoft CEO Satya Nadella. Mustafa Suleyman remained at MAI but shifted his focus to overseeing the MAI Superintelligence division. This reorganization clarified the separation between product-focused AI development and long-term research into superintelligence.

## Research and Collaborations

MAI's research efforts build on a rich ecosystem of academic and industrial partnerships. Microsoft has a long history of collaboration with institutions like [MIT CSAIL](https://www.wikiprompt.org/wiki/mit-csail), [Stanford AI Lab](https://www.wikiprompt.org/wiki/stanford-ai-lab), [University of Toronto](https://www.wikiprompt.org/wiki/university-of-toronto), and [Carnegie Mellon University](https://www.wikiprompt.org/wiki/carnegie-mellon-university). The division also works with cloud infrastructure providers, including its own [Azure](https://www.wikiprompt.org/wiki/azure) platform, which offers [AWS](https://www.wikiprompt.org/wiki/amazon-web-services)-like services but is a separate entity.

Key research areas include [machine learning](https://www.wikiprompt.org/wiki/machine-learning), [deep learning](https://www.wikiprompt.org/wiki/deep-learning), and [neural networks](https://www.wikiprompt.org/wiki/neural-network), with contributions from notable researchers such as [Karén Simonyan](https://www.wikiprompt.org/wiki/karen-simonyan), who is known for work on convolutional networks, and [Jakob Uszkoreit](https://www.wikiprompt.org/wiki/jakob-uszkoreit), a co-inventor of the transformer architecture. MAI also draws on advances from [OpenAI](https://www.wikiprompt.org/wiki/openai), [Anthropic](https://www.wikiprompt.org/wiki/anthropic), and [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind) in the broader AI landscape.

## Products and Impact

MAI's flagship product is Microsoft Copilot, an AI assistant integrated across Windows, Edge, Bing, and Microsoft 365. Copilot leverages [large language models](https://www.wikiprompt.org/wiki/large-language-model) to provide conversational assistance, code generation, and content creation. The division also oversees GroupMe, which was integrated with Copilot in 2025, and continues to develop tools for healthcare and scientific discovery.

The in-house models, such as MAI-Voice-1 and MAI-Image-1, are designed to compete with offerings from other tech giants like [Apple](https://www.wikiprompt.org/wiki/apple), [Samsung](https://www.wikiprompt.org/wiki/samsung-electronics), and [Google Cloud](https://www.wikiprompt.org/wiki/google-cloud). By developing proprietary models, MAI aims to reduce costs, improve integration, and maintain control over its AI stack.

## Future Directions

As of 2026, MAI continues to expand its research and product portfolio. The superintelligence division represents a long-term bet on achieving advanced AI capabilities, with applications in companionship, healthcare, and clean energy. The Copilot division focuses on delivering practical AI tools to consumers and businesses. With offices across the globe and a strong research culture, MAI is positioned as a major player in the global AI race, competing with [OpenAI](https://www.wikiprompt.org/wiki/openai), [Anthropic](https://www.wikiprompt.org/wiki/anthropic), and [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind).

MAI's trajectory suggests a continued emphasis on in-house model development, strategic partnerships, and integration of AI into everyday Microsoft products. The division's leadership, including Mustafa Suleyman and Karén Simonyan, brings deep expertise from leading AI organizations, and the company's investment in research hubs like London underscores its commitment to innovation.

---
Source: https://www.wikiprompt.org/wiki/microsoft-ai
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-08T15:32:27.441103+00:00