Tracy Hresko Pearl (U Oklahoma College Law) has posted “Governance in the Absence of Government” on SSRN. Here is the abstract:
Artificial intelligence (AI) is advancing at an unprecedented rate, with generative AI systems demonstrating increasing autonomy and influence over social, economic, and political systems. While AI has the potential to transform society for the better by automating complex tasks and accelerating innovation, it also has the potential to inflict great harm. Both industry and government officials have raised significant concerns about the use of AI to spread misinformation, the potential for AI to create widespread job displacement, and the existential threats associated with highly autonomous AI systems operating outside of human control. As a result, there is widespread agreement among industry leaders, government officials, and citizen groups that some form of AI oversight is necessary.
However, AI presents a formidable challenge to traditional lawmaking and regulatory frameworks in the United States. There are six key barriers to effective government regulation of AI, in particular: (1) the rapid pace of technological advancement outstripping legislative processes, (2) a lack of AI expertise among policymakers, (3) regulatory capture, (4) political gridlock, (5) outdated regulatory structures, and (6) the complexity of AI itself. Given these constraints and the cross-border nature of the AI industry, it is highly unlikely that the U.S. government will be able to implement effective AI regulations in time to meaningfully mitigate risks, necessitating the exploration of alternatives to government action.
Soft law, which includes industry standards, voluntary compliance mechanisms, and multi-stakeholder initiatives, offers one viable alternative to government regulation and has successfully shaped governance in other industries in which formal regulation has lagged. Soft law is a significantly more agile tool for governance of emerging technologies, allowing for rapid iteration and cross-border cooperation without the delays inherent in formal legislation. It encourages dialogue and collaboration among a broad range of stakeholders and can promote innovation by reducing the likelihood of a patchwork of conflicting laws and regulations arising across jurisdictions. It is not, however, a panacea. Soft law’s non-binding nature can undermine its effectiveness, and powerful industry players can sometimes dominate soft law institutions. However, when developed thoughtfully, soft law initiatives can be a remarkably effective form of governance when governments are otherwise unwilling or incapable of regulating an industry effectively.
Given the complexity and cross-border nature of AI, an International Council on AI Risk (ICAR), a multi-stakeholder body dedicated to AI risk mitigation and global standard-setting, should be established as swiftly as possible. An ICAR could bring together governments, industry leaders, academics, and citizen groups. It could also coordinate AI safety measures, establish compliance mechanisms, and develop adaptable regulatory frameworks. Unlike rigid government structures, an ICAR could operate dynamically, responding to AI’s rapid evolution while providing enforceable soft law mechanisms that would encourage responsible AI development. While such an institution would undoubtedly face challenges in defining its mission and scope, ensuring industry compliance, and avoiding bureaucratic stagnation, these challenges are surmountable through strong multi-stakeholder participation and carefully designed governance structures. Indeed, given the inadequacy of government-led regulation and the high stakes of AI development, an ICAR represents the most promising governance model for ensuring AI safety. By leveraging international cooperation, industry incentives, and adaptive oversight, an ICAR could offer a pragmatic and scalable alternative to traditional regulatory approaches, addressing AI’s growing risks in the absence of effective government action.