Artificial intelligence is rapidly reshaping the global economy. Governments and corporations increasingly portray AI as essential to economic competitiveness, scientific advancement and national security. Yet alongside its promise comes a growing recognition that AI systems can amplify bias, erode privacy, spread misinformation and deepen existing inequalities if not governed responsibly.

In response, governments, international organizations and technology companies have developed frameworks designed to ensure that AI remains ethical, accountable and worthy of public trust. Across these frameworks, several common principles emerge: transparency, accountability, fairness, safety, privacy protection and meaningful human oversight.

Together, they seek to answer a central question: What makes AI trustworthy?

Related: Here’s what communities can do when data centers arrive

Yet a growing contradiction sits at the heart of the AI economy.

While technology companies champion “responsible AI,” many of the massive data centers powering AI systems are being developed through processes that communities increasingly describe as opaque, exclusionary and insufficiently accountable. Across Minnesota, residents have raised concerns not only about water consumption, electricity demand and environmental impacts, but also about how decisions surrounding large-scale data center development are being made.

This raises an uncomfortable but necessary question: Can AI truly be considered trustworthy if the infrastructure powering it is developed through secrecy, weak accountability or procedural unfairness?

A useful analogy comes from American constitutional law: the doctrine known as the “fruit of the poisonous tree.” The doctrine generally holds that evidence obtained through unconstitutional conduct is tainted and may be excluded from court proceedings. The underlying principle is simple: Flawed processes can compromise the legitimacy of outcomes.

As a governance analogy, the doctrine offers a useful lens for evaluating AI infrastructure. If data centers are approved through opaque decision-making, inadequate public participation, weak disclosure practices or limited accountability, then the legitimacy of the AI systems they support may also be called into question.

Minnesota offers an important case study.

Driven largely by growing demand for AI computing power, the state has experienced a surge of proposed hyperscale data centers. Communities including Farmington, Monticello, Hermantown, North Mankato, Eagan and Worthington have found themselves at the center of debates over projects capable of reshaping local energy systems, water resources and land-use planning.

The scale of these facilities is difficult to ignore. Hyperscale data centers require enormous quantities of electricity and, depending on cooling technologies, can consume millions of gallons of water annually. As more facilities are proposed, questions naturally arise about cumulative impacts on Minnesota’s energy infrastructure, climate goals and natural resources.

Yet opposition to data centers is not simply opposition to technology.

Many community concerns focus on governance. Residents have questioned whether public officials and developers have provided sufficient information about project scope, energy demand, water consumption and long-term community impacts before critical decisions are made. Reports that some negotiations have involved nondisclosure agreements have further fueled concerns about transparency and public accountability.

At their core, community concerns generally fall into five categories.

First, transparency. Residents want clear and accessible information about water use, energy demand, environmental impacts and public subsidies.

Second, accountability. Communities expect developers and public officials to follow lawful procedures and provide accurate information throughout the permitting and approval process.

Third, fairness. Residents want a meaningful opportunity to participate in decisions that may affect their neighborhoods, property values, environmental quality and utility costs.

Fourth, sustainability. The substantial energy and water requirements of large data centers raise legitimate questions about long-term environmental impacts and grid resilience.

Finally, public trust. When communities perceive secrecy or exclusion, confidence in both institutions and technology companies can erode.

Ironically, these are many of the same principles embedded in leading AI governance frameworks.

The UNESCO Recommendation on the Ethics of Artificial Intelligence emphasizes human rights, environmental sustainability and democratic participation. The National Institute of Standards and Technology’s AI Risk Management Framework identifies trustworthiness as a foundational objective. The European Union’s AI Act prioritizes transparency, accountability and protection of fundamental rights.

Yet much of the AI governance conversation remains focused on the design, development and deployment of AI models. Those issues are important, but trustworthy AI cannot be reduced to algorithms alone. The physical infrastructure that powers AI deserves scrutiny as well.

Responsible AI should also require responsible infrastructure governance: transparent environmental review, meaningful public participation, honest disclosure of environmental impacts, enforceable accountability mechanisms and respect for democratic decision-making.

Related: Why local officials in Minnesota are signing nondisclosure agreements

Technology companies often argue that public trust is essential to successful AI adoption. They are correct. But trust is not built solely through ethics statements, governance frameworks or corporate commitments. Trust is earned through transparent and accountable conduct across the entire AI ecosystem, including the data centers that make modern AI possible.

The future of artificial intelligence will not be determined solely in Silicon Valley boardrooms or international regulatory agencies. It will also be shaped in communities across Minnesota and beyond, where residents are increasingly demanding a voice in decisions about the infrastructure behind the AI revolution.

The lesson underlying the “fruit of the poisonous tree” doctrine remains relevant beyond the courtroom: legitimacy matters. Process matters. How something is built matters as much as what is built.

If the infrastructure powering AI is developed through secrecy, exclusion or governance failures, public confidence in trustworthy AI will become increasingly difficult to sustain.

There can be no fully trustworthy AI without trustworthy infrastructure

Eric Ini is a community leader and policy advocate with more than 10 years of experience advancing environmental justice, human rights and governance across Africa, Asia and the United States. He most recently served as chief equity and partnership officer at the Minnesota Center for Environmental Advocacy (MCEA), where he led equity, community engagement, and partnership initiatives. Eric writes on issues at the intersection of environmental justice, access to justice and AI governance.

Related