AI Governance at the Frontier: Unpacking Foundational Assumptions

Executive Summary 1

Introduction 3

Report Scope 5

Guiding Questions 5

Selection Criteria 6

Limitations 6

Proposal Case Studies 7
    Industry: OpenAI’s Approach to Frontier Risk 7
    Civil Society: AI Now Institute’s Zero Trust AI Governance 9
    Academia: Frontier AI Regulation: Managing Emerging Risks to Public Safety 11
    State Government: California’s Safe and Secure Innovation for Frontier Artificial Intelligence Models Act 13
    Federal Government: Framework to Mitigate AI-Enabled Extreme Risks 17

Analysis 19
    Deriving Assumptions 19
    Shared Assumptions 20
    What Risks Are Important to Mitigate And Who Should Have Primary Oversight of Frontier AI 21
    Who Is Delegated Tasks and Able to Play a Role 22
    Would the Proposed Mechanisms or Tools Actually Achieve the Proposal’s Objectives 24

Unique Assumptions 30

Discussion 32

Conclusion 34

Authors 35

Acknowledgments 35

Appendix A: Assumption Interdependencies 36

Endnotes 37