Compute-Based Regulations

Summary

This paper from the Centre for the Governance of AI examines regulatory approaches that use computational resources (compute) as a lever for governing AI development and deployment. It analyzes how tracking, limiting, or monitoring compute usage could enable policymakers to oversee frontier AI systems and enforce safety standards. The paper evaluates the feasibility, advantages, and limitations of compute-based regulatory frameworks.

Key Points

• •Compute is a measurable and controllable input to AI development, making it a tractable point of regulatory intervention.
• •Proposes frameworks for using compute thresholds to trigger reporting requirements, safety evaluations, or licensing obligations.
• •Examines hardware supply chain controls and cloud computing oversight as complementary enforcement mechanisms.
• •Discusses challenges including jurisdictional limitations, algorithmic efficiency improvements that reduce compute needs, and risks of over-regulation.
• •Situates compute-based regulation within broader AI governance strategies, noting it complements but does not replace other oversight approaches.