MWC 2023: Microsoft Introduces 'AI Access Principles' to Address Competition Concerns with OpenAI

Following Microsoft’s recent investment and collaboration with French startup Mistral AI, the company aims to shift the narrative around its perceived monopolistic practices stemming from its significant stake in OpenAI. At the Mobile World Congress on Monday, Microsoft unveiled its "AI Access Principles," an eleven-point framework designed to "govern how we will operate our AI data center infrastructure and other critical AI assets globally."

The principles address various key areas, including the creation of an app store that allows businesses to select different Large Language Models (LLMs) and AI products. Additionally, Microsoft commits to keeping proprietary data separate from its training models and enabling customers the flexibility to switch cloud providers or services as needed. The framework emphasizes cybersecurity for AI services, environmentally sustainable construction of data centers and infrastructure, and investments in education.

Brad Smith, Microsoft’s President and Vice Chair, announced the framework during his keynote at Mobile World Congress in Barcelona. While this declaration indicates Microsoft’s willingness to engage with stakeholders, it is worth noting that Smith delivered the announcement without allowing for any follow-up questions.

This announcement arrives amidst increasing regulatory scrutiny concerning Microsoft’s $13 billion investment in OpenAI, which has granted them a 49% share in the leading generative AI service provider. The European competition authority has indicated it is reviewing whether this investment aligns with antitrust regulations.

The AI Access Principles specifically address how third parties might utilize Microsoft’s platforms and services for AI product development, an essential area of growth the company seeks to expand not only with the Mobile World Congress attendees but also a broader spectrum of industries.

"If they are training a model on our infrastructure or deploying it, we recognize their data belongs to them; we will not access or use it to compete against those relying on our infrastructure," Smith stated.

It is important to note that these AI Access Principles are not legally binding and lack specific mechanisms for monitoring compliance. However, they signal Microsoft's commitment to fostering competition, which may be leveraged during any formal regulatory investigations.

Smith emphasized the scale of Microsoft’s AI operations, announcing, "As of today, we have nearly 1,600 models running in our data centers, 1,500 of which are open source models," showcasing the company's commitment to both proprietary and open-source innovations across companies of all sizes.

By publicly sharing these principles, Microsoft creates a declaration that competitors, the public, and regulators can reference to evaluate whether the company lives up to its stated commitments.

AI, Europe, Government & Policy, MWC, MWC 2024, Spain