Practical Principles for AI Cost and Compute Accounting

TL;DR

This paper proposes seven principles for AI cost and compute accounting standards to improve regulatory effectiveness, prevent strategic gaming, and promote responsible AI development across different organizations and regions.

Contribution

It introduces a set of practical principles to guide the design of AI cost and compute accounting standards addressing current technical ambiguities.

Findings

Seven principles to guide AI cost and compute accounting standards.

Principles aim to reduce strategic gaming and disincentives.

Facilitates consistent implementation across jurisdictions.

Abstract

Policymakers increasingly use development cost and compute as proxies for AI capabilities and risks. Recent laws have introduced regulatory requirements for models or developers that are contingent on specific thresholds. However, technical ambiguities in how to perform this accounting create loopholes that can undermine regulatory effectiveness. We propose seven principles for designing AI cost and compute accounting standards that (1) reduce opportunities for strategic gaming, (2) avoid disincentivizing responsible risk mitigation, and (3) enable consistent implementation across companies and jurisdictions.