Public Utility Regulation
Common carrier obligations, universal access requirements, and price controls to prevent monopolistic capture of essential AI services.
What it is:
Public utility regulation would subject frontier AI systems and compute infrastructure to government oversight similar to electricity grids, telecommunications networks, or water systems. Under this framework, AI providers would face obligations including non-discriminatory access (serving all customers on equal terms), regulated pricing (preventing monopolistic rent extraction), reliability standards (ensuring consistent service availability), and transparency requirements (disclosing capabilities and limitations). The underlying premise is that some technologies become so essential to economic participation that market provision alone produces unacceptable outcomes — monopoly pricing, exclusionary access, or service instability that cascades through dependent systems. Public utility regulation establishes ex ante obligations rather than waiting to address harms after they occur.
If frontier AI capabilities concentrate in a small number of providers, those providers gain leverage over every downstream business and worker that depends on their services — the power to set prices, terms of access, and acceptable use. Utility regulation would constrain this leverage by requiring providers to serve all customers on comparable terms and submit pricing to regulatory review. This has distributional implications: regulated pricing prevents AI providers from capturing the full surplus their systems generate, leaving more value with businesses and workers who use AI as an input. Universal access requirements could also ensure that AI capabilities diffuse broadly rather than remaining concentrated among firms and individuals who can afford premium pricing, potentially reducing the gap between AI's beneficiaries and those displaced by it.
The challenge:
The main difficulty is defining what counts as essential. While electricity and water have clearly defined boundaries, AI capabilities are heterogeneous, rapidly evolving, and bundled in complex ways that resist standardization. There is also a tension between the stability that utility regulation provides and the dynamism that characterizes AI development, since regulation optimized for consistency may slow the capability gains that benefit users. Utility-style regulation could also entrench incumbents by codifying their current systems as the regulatory standard.
Recommended Reading:
Real-world precedents:
The Interstate Commerce Act of 1887 made railroads the first industry subject to federal regulation, requiring "just and reasonable" rates and prohibiting discriminatory pricing.
The Federal Communications Commission regulates Internet Service Providers under Title II of the Communications Act as common carriers, imposing non-discriminatory service obligations.