2023 UK AI Safety Summit Response (Microsoft)

• Authority: Private-sector companies
• Collections: Corporate policies and commitments
• Most recent activity: Proposed
• Most recent activity date: 2023-10-26

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• Exposure to toxic content
• AI system security vulnerabilities
• Disinformation, surveillance, and influence at scale
• Cyberattacks, weapon development or use
• Fraud, scams, and targeted manipulation
• Competitive dynamics
• Governance failure
• AI pursuing its own goals
• AI possessing dangerous capabilities
• Lack of capability or robustness
• Lack of transparency or interpretability

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Compromise of privacy
• False or misleading information
• Overreliance and unsafe use

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 10-12 subdomains, with strong focus on AI system security (2.2), malicious actors and misuse (4.1, 4.2, 4.3), competitive dynamics (6.4), governance structures (6.5), and AI safety failures (7.1, 7.2, 7.3, 7.4). Coverage is concentrated in security, misuse prevention, AI safety, and governance domains, with minimal coverage of discrimination, privacy, misinformation, and socioeconomic impacts.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Information
• Scientific Research & Development Services

All other sectors had no coverage in this document.

Sectors Summary

This is an internal corporate policy document from Microsoft, an AI developer and technology company. The sectors governed are primarily Information (where Microsoft operates as a technology and AI company) and Scientific Research and Development Services (through Microsoft Research's AI research activities). The document does not regulate external sectors but describes Microsoft's internal governance of AI development and deployment across its own operations.

Legislative Status

Other

Legislative Status Summary

This is an internal corporate policy document describing Microsoft's voluntary commitments and responsible AI practices. It references voluntary commitments made at the White House convening and describes internal governance structures, standards, and processes that Microsoft has implemented.

Actors

Proposer

• Entity Types: AI_Developer
• Entities: Microsoft; OpenAI
• Summary: Microsoft is the primary proposer of this governance framework, describing its own responsible AI policies, standards, and practices. The document is Microsoft's response to the UK Government's AI Safety Policies Request, outlining voluntary commitments and internal governance structures.

Target

• Entity Types: AI_Developer
• Entities: Microsoft product teams; Microsoft engineering teams; OpenAI
• Summary: The governance framework applies to Microsoft's internal product teams, engineering teams, and development processes. It also applies to the Microsoft-OpenAI collaboration on frontier model development and deployment.

Enforcer

• Entity Types: AI_Developer
• Entities: Responsible AI Council; Microsoft-OpenAI Deployment Safety Board (DSB); Office of Responsible AI; Microsoft Security Response Center (MSRC); Digital Security & Resilience team
• Summary: Microsoft's internal governance bodies enforce the responsible AI policies, including the Responsible AI Council co-chaired by Brad Smith and Kevin Scott, the joint Deployment Safety Board with OpenAI, and various security and compliance teams.

Monitor

• Entity Types: AI_Developer
• Entities: Office of Responsible AI; AI Red Team; Microsoft Security Response Center (MSRC); Responsible AI Division Leads and Champs; Microsoft Threat Intelligence
• Summary: Multiple internal teams monitor compliance and implementation, including the Office of Responsible AI, the AI Red Team for security testing, and various monitoring and threat intelligence teams that track ongoing performance and risks.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Plan and Design
• Collect and Process Data
• Build and Use Model
• Verify and Validate
• Deploy
• Operate and Monitor

AI Lifecycle Summary

The document comprehensively covers all stages of the AI lifecycle, with particularly strong emphasis on Build and Use Model, Verify and Validate, Deploy, and Operate and Monitor stages. It describes detailed processes for model development, evaluation, testing, deployment review, and ongoing monitoring.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• Foundation Models
• Frontier AI
• AI Systems
• Generative AI

Technical Scope Summary

The document explicitly covers AI models, AI systems, and frontier AI with detailed discussion. It mentions foundation models and generative AI systems extensively. The document does not explicitly define compute thresholds or discuss general purpose AI, task-specific AI, predictive AI, or open-weight models in detail, though it references model capabilities broadly.