AI is free for most users, but its true costs are landing on specific communities through higher utility bills, depleted water supplies, and unpaid creative labor. Three distinct approaches emerging across the globe reveal how different regions are tackling the question of who should bear the financial burden as artificial intelligence becomes embedded in everyday life (Sources 1, 2, 3).

Why Are Communities Paying for AI They Don't Use?

The fundamental inequity is straightforward: AI's benefits are broad and diffuse, but its consequences are concentrated and localized. Neighborhoods hosting data centers experience real, measurable harm. In Ohio, residents served by AEP Ohio, the power company serving the state's data center corridor, were told to expect monthly electricity bills to climb by $27 due in part to soaring energy demands from nearby tech facilities. Residents experience utility strain and rising costs regardless of whether they personally use AI services.

Water depletion presents another crisis. Data centers require enormous quantities of water for cooling, and neighborhoods that host these facilities have experienced supply interruptions. Beyond infrastructure strain, creative professionals face a different kind of loss: their work is extracted for AI training without consent or compensation. Most artists, writers, and musicians never agreed to have their creations used to train profitable AI systems, yet they now compete with AI tools trained on their own voices.

How Can Taxation Address AI's Externalities?

Researchers at Carnegie Mellon University's Department of Engineering and Public Policy have proposed a framework using targeted taxation to mitigate these costs. According to their analysis, taxation can work in three complementary ways:

• Corrective Function: Taxes on water and electricity consumption create price signals that discourage overconsumption by AI data centers, making resource-intensive operations more expensive.
• Redistributive Function: Tax revenue can be directed toward restorative or remedial action in affected communities, compensating residents and creative professionals for harms incurred.
• Regulatory Capacity: Tax revenue can fund AI oversight infrastructure, including monitoring and enforcement of regulations that currently lack adequate resources.

"AI taxation deserves closer consideration by policymakers, as it could add the missing public-finance piece of AI governance: how to pay for the needs an AI economy creates," said Juliette Faivre, an incoming Ph.D. student in Engineering and Public Policy at Carnegie Mellon.

Juliette Faivre, Incoming Ph.D. Student, Engineering and Public Policy, Carnegie Mellon University

The appeal of taxation lies in its proven track record. Mitigation taxes, often called Pigouvian taxes, have successfully offset societal costs of private industry goods and behaviors for decades in the United States. Unlike complex regulatory frameworks that require untested evaluation infrastructure, tax compliance mechanisms already exist in most businesses, allowing swift deployment. Additionally, in the U.S., tax laws can be passed through budget reconciliation, bypassing typical congressional gridlock.

The proposal is no longer purely theoretical. Senate proposals addressing AI-driven harms have been introduced, and some industry leaders have recently supported AI taxation discussions.

What Are the Limitations of an AI Tax Approach?

Researchers acknowledge that taxation is not a comprehensive solution. Firms may attempt to avoid taxes by relocating to jurisdictions with lower taxation or weaker regulatory enforcement. Taxation may also be inappropriate for certain harms, such as privacy violations and discrimination, which require different legal remedies. There is also a persistent risk that taxation could chill innovation and threaten global competitiveness. For these reasons, experts emphasize that taxation should complement other AI governance tools rather than serve as a standalone fix.

How Are Other Countries Building AI Governance Frameworks?

While the United States debates taxation mechanisms, other nations are constructing broader governance structures. Burkina Faso is taking a methodical approach by educating its lawmakers about AI before drafting regulations. On July 15, the Permanent Secretariat for Innovation and Monitoring of Emerging Digital Technologies (SPIVTEN), under the Ministry of Digital Transition, held a training session in Ouagadougou for approximately 30 members of the Legislative Assembly of the People (ALP).

The program introduced lawmakers to AI fundamentals, practical applications, and policy challenges. Discussions covered digital sovereignty, data governance, and cybersecurity. The goal is to equip legislators with sufficient understanding to draft laws governing AI use while protecting national interests. This initiative follows Burkina Faso's adoption of a national AI roadmap in August 2025 and subsequent public-sector training programs. Since then, the government has expanded AI training across multiple levels, including sessions for senior officials in April 2026 and Prime Minister's Office training in May and July on AI opportunities, risks, and ethical use.

Australia, meanwhile, has announced a more comprehensive national strategy. Prime Minister Anthony Albanese unveiled the country's National AI Plan, positioning AI as a strategic growth industry backed by targeted public funding and new governance structures, including an Office of AI under the prime minister and cabinet. The plan consolidates over A$460 million in existing AI-related government funding into a single industrial strategy aimed at growing a "world-class AI ecosystem".

A distinctive element of Australia's approach is its focus on creator compensation. Albanese declared that creators of books, music, art, and news "should retain control of the price and value of their work" when used to train AI, stating that "anything less is theft." Forthcoming national AI standards will include protections for creative professionals whose works are ingested into AI models, though enforcement mechanisms remain to be detailed.

Albanese

Australia also addresses infrastructure sustainability directly. Large AI data centers will be required to generate as much power as they consume and meet stringent water-efficiency expectations, effectively imposing energy and water guardrails tied to AI growth. Government-developed national data-center principles will set expectations on environmental and community impacts.

How Do These Governance Models Compare?

The three approaches reflect different governance philosophies. The U.S. taxation model leverages existing financial mechanisms to redistribute costs and fund oversight. Burkina Faso's approach prioritizes legislative capacity-building and institutional readiness before formal regulation. Australia's strategy combines industrial policy, infrastructure standards, creator protection, and centralized coordination without relying on EU-style prescriptive regulation.

Australia's model differs from the European Union's AI Act, which uses prescriptive, horizontal, risk-based regulation with extraterritorial obligations. Instead, Australia employs a "strategy-plus-standards" approach that is largely technology-neutral, focusing on specific applications and infrastructure rather than broad regulatory categories. For companies serving both markets, governance calibrated to the EU AI Act will generally exceed Australian requirements, with Australia adding distinctive infrastructure sustainability and creator protection elements.

Despite Australia's claims of a "world-leading" framework, the policy approach resembles state-level AI strategies in the United States more closely than it represents a fundamentally novel governance model. Centralized funding and control of industrial policy direction are occurring at state level as AI infrastructure and applications are deployed across the U.S. .

What Comes Next for AI Governance?

These three approaches suggest that AI governance is moving beyond voluntary principles toward binding mechanisms. Whether through taxation, legislative education, or infrastructure standards, policymakers are recognizing that AI's costs cannot remain externalized indefinitely. The challenge ahead is designing frameworks that protect affected communities and creative professionals while maintaining innovation incentives and global competitiveness. As more countries develop their own AI strategies, the question of who pays for AI's true costs will likely become central to international trade negotiations and regulatory harmonization efforts (Sources 1, 2, 3).