What happened
RAND published this research report on June 24, 2026, drawing on a targeted literature review, 14 semi-structured practitioner interviews, and five RAND-designed tabletop exercises to develop a structured framework for how non-governmental actors can govern transformative AI risks. The framework identifies three roles for business and civil society: managing technical and operational risks within development and deployment; shaping incentives for safety through market and network mechanisms; and supporting social stability and public trust during AI-related transitions. A clear finding across all three research methods is that 'business and civil society actors are well suited to contribute to managing and mitigating transformative AI risks because of their positions in the AI development and deployment life cycles, as well as in society.' Under each role the report specifies minimum-baseline actions feasible under existing legal and market conditions, and distinguishes these from coordinated next steps — including formal AI risk functions with escalation routes, safety criteria in investor due diligence and procurement, and workforce transition planning by employers and unions.
Why it matters
As government regulatory cycles lag AI capability growth and frontier development remains concentrated in private firms, this report gives executives, boards, and institutional investors a concrete, evidence-based playbook for taking ownership of AI governance responsibilities that government alone cannot discharge in time.
Action needed
Map the three-role framework (manage technical risk, shape market incentives, support social stability) against your organisation's current AI governance posture and identify which minimum-baseline actions — named executive ownership, AI-use inventory, human-override procedures — are not yet in place.