Our Pledge to Enhance AI Safety, Security, and Reliability

Microsoft Endorses White House's AI Guidelines, Unveils Additional PledgesToday, Microsoft is affirming its support for new voluntary guidelines put forth by the Biden-Harris administration aimed at ensuring advanced AI systems are secure, reliable, and trustworthy. Microsoft is not only endorsing the guidelines outlined by President Biden but also taking independent steps that bolster these essential objectives, thereby broadening its commitment to responsible AI deployment along with other leaders in the sector. The swift actions proposed by the White House lay the groundwork to ensure AI's potential benefits outweigh its risks.
We are encouraged by the President's initiative in rallying the tech community to define practical measures that will enhance AI's safety, security, and utility for everyone. Rooted in the core values of safety, security, and trust, these voluntary guidelines tackle the challenges posed by sophisticated AI technologies and encourage the adoption of specific strategies—such as red-team testing and releasing transparency reports. These efforts are designed to advance the entire field and build on significant existing U.S. initiatives, including:

• Complementing NIST AI Risk Management Framework and AI Bill of Rights
  

These efforts complement stringent regulations already in place for high-risk applications in Europe and beyond. We anticipate widespread industry adoption of these practices and their inclusion in broader international discussions about forming an effective global AI conduct code.
The Call for Responsible AI: Industry Collaboration and Safety MeasuresMicrosoft's additional pledges are aimed at enhancing the ecosystem and actualizing the principles of safety, security, and trust. From supporting trials of the National AI Research Resource to promoting the creation of a national registry for high-risk AI systems, we believe these actions will promote greater transparency and accountability.
We are also committed to a widespread application of the NIST AI Risk Management Framework and to adopting cybersecurity measures specifically designed to address unique AI threats, which we expect will lead to more dependable AI systems that will benefit our customers and society at large. Details on Microsoft’s commitments can be found here.
Building Trustworthy AI: Specific Strategies and Existing InitiativesDeveloping commitments of this nature and implementing them at Microsoft requires a collective effort.
Microsoft Backs Biden-Harris Administration's Voluntary AI GuidelinesI extend my gratitude to Kevin Scott, Microsoft’s Chief Technology Officer, with whom I co-lead our responsible AI initiative, and to Natasha Crampton, Sarah Bird, Eric Horvitz, Hanna Wallach, and Ece Kamar, for their pivotal roles in our responsible AI framework.
As emphasized by the White House’s voluntary guidelines, it is crucial to keep human interests at the forefront of AI development.

• Red Team Testing and Transparency Reports for Secure AI
  

Microsoft's Commitment to Responsible AI: Enhancing the EcosystemI am thankful for the robust leadership team at Microsoft that helps us fulfill our pledges and continue to refine the program we have been shaping over the past seven years. Establishing ethical standards early in the evolution of this burgeoning technology not only secures its safety, security, and trustworthiness but also maximizes AI’s beneficial impacts for communities across the U.S. and globally.

• National AI Research Resource and National Registry for High-Risk AI Systems
  
• Widespread Adoption of NIST Framework and Cybersecurity Measures for AI
  

Human-Centric AI Development: A Collaborative Effort at Microsoft

• Microsoft's Responsible AI Initiative: A Collaborative Effort
  
• The Importance of Ethical Standards in AI Development