The AI Governance Vacuum: Why the US Must Lead on Standards

The recent calls for a standards body to oversee artificial general intelligence (AGI) and mitigate national security risks have highlighted the urgent need for global cooperation in regulating this rapidly advancing field. At the forefront of these discussions is Google DeepMind’s chief Demis Hassabis, who has been vocal about the US taking the lead in developing an AI framework.

Hassabis’ proposal for a public-private partnership overseen by the federal government recognizes the US’s unique position to shape the global conversation on AI governance. With its economic and technical standing, the US can establish a robust framework that balances innovation with safety.

The challenge lies in reconciling the complex relationship between public and private sectors. Disputes over export controls and model releases have shown that regulating leading AI models is increasingly contentious. Companies like Anthropic and OpenAI face restrictions from the US government, while Chinese companies are rapidly gaining traction in the global market.

Tensions between the two superpowers are palpable, with lawmakers considering how to curb the adoption of Chinese AI models by homegrown companies. This raises serious concerns about national security implications of relying on foreign technologies that may be vulnerable to cyber threats or even malicious intent.

Hassabis’ suggestion for a Standards Body modelled on FINRA’s self-regulatory organization is an intriguing one. By including independent technical experts and open-source representatives, such a body could provide a platform for global collaboration and ensure AI development aligns with societal values.

The proposed review process, where frontier labs share models with the body 30 days before release, is also crucial. This allows for early identification of potential risks and vulnerabilities, preventing catastrophic failures that might have far-reaching consequences for national security.

The US must consider its historical context in AI governance. The development of AGI has been compared to nuclear power – a double-edged sword requiring careful management to harness benefits while mitigating risks. International agreements on nuclear non-proliferation demonstrate the need for similar efforts to regulate AI.

In the coming months, policymakers, industry leaders, and technologists must work together to establish a robust framework for innovation with safety. If the US fails to take charge of AI standards, other nations will fill the vacuum. China’s willingness to invest heavily in AI research and development makes this a pressing concern.