Microsoft outlined a series of principles for its development and deployment of artificial intelligence, including a pledge to partner with a “broad array” of companies and other players in the industry.
The company announced a multi-year partnership and an investment in French startup Mistral AI as an example of its approach. Mistral AI will be able to train and deploy its AI models via Microsoft’s data centers, and Microsoft will make Mistral AI’s models available to customers through its Azure cloud service.
[Update: Microsoft invested 15 million euros in Mistral AI, or about $16.2 million at current exchange rates, “which would convert into equity in Mistral’s next funding round,” a Microsoft spokesperson said via email.]
“We’re focused not just on proprietary software, not just on our partnership with OpenAI, as critical as it is, but with a wide number of companies,” said Brad Smith, the Microsoft president and vice chair, outlining the company’s AI access principles during a keynote address at Mobile World Congress in Barcelona on Monday.
The principles include commitments to provide broad access and support for AI developers; ensure choice and fairness across the emerging AI economy; operate in a socially responsible manner on issues of privacy, safety, and security; support strong physical and cybersecurity; promote people-centered AI; and invest in AI skilling initiatives.
Smith said the AI principles build on Microsoft’s “Windows Principles,” which the company outlined in 2006 in an effort to address and move on from longstanding antitrust challenges over its dominant PC operating system.
The New York Times on Sunday detailed the evolution of Microsoft’s approach to public policy issues, describing the company’s strategy of “self-interested enlightenment, taking positions that appeared calculated to highlight the ways it had reformed itself and to deflect scrutiny toward competitors,” as the story described it.
Microsoft recently announced $5.6 billion in AI data center investments in Europe.
The company’s AI principles comes against a backdrop of growing regulatory scrutiny, including the FTC’s announcement Jan. 25 that it “will scrutinize corporate partnerships and investments with AI providers to build a better internal understanding of these relationships and their impact on the competitive landscape.”
The FTC said at the time that it sent compulsory requests for information to Google parent Alphabet; Microsoft and its partner OpenAI; and Amazon and its partner Anthropic. (Amazon separately announced last week that Mistral’s AI models will also be coming to the AWS Bedrock service.)
In his speech and a related blog post, Smith sought to differentiate Microsoft from some of its biggest competitors, saying that Microsoft doesn’t require developers to use its app store for distribution like Apple does, and doesn’t operate a vertically integrated business like Google does.
“Today, only one company is vertically integrated in a manner that includes every AI layer from chips to a thriving mobile app store,” Smith wrote in his blog post, referring to Google without naming the company.
GeekWire has contacted Google for comment on Microsoft’s announcement.
When the FTC announced its AI probe in January, a Google Cloud spokesperson said via email, “We hope the FTC’s study will shine a bright light on companies that don’t offer the openness of Google Cloud or have a long history of locking-in customers – and who are bringing that same approach to AI services.”
Update: In an interview with Reuters, Google Cloud Vice President Amit Zavery said the company is closely monitoring Microsoft’s negotiations to resolve a complaint in Europe from CISPE, cloud infrastructure services providers who say Microsoft is leveraging its dominance in productivity software to its benefit in the cloud.
As part of Microsoft’s AI principles, Smith said that the company is committed to giving customers the choice to leave if they want to move their model or data and to another cloud. He said it is Microsoft’s responsibility to enable people to easily migrate to other cloud providers.
From Smith’s speech and blog post, part of his goal in presenting the AI principles was to encourage a more nuanced understanding of the different approaches needed at each layer of the AI stack.
“While one company currently produces and supplies most of the GPUs being used for AI today, as one moves incrementally up the stack, the number of participants expands, Smith wrote, referring to Nvidia. “And each layer enables and facilitates innovation and competition in the layers above. In multiple ways, to succeed, participants at every layer of the technology stack need to move forward together.”
