Presidents, Kings, and the Fight for Reality: Why Democracy Needs Both Law and Trustworthy AI
When Authority Becomes Indistinguishable from Power
Imagine two very different moments in modern history. In one, a president claims absolute power and refuses to hand over evidence, just as Nixon did until the Supreme Court compelled him to release his tapes. In another, a deepfake video of a political candidate floods millions of feeds within hours, convincing citizens of something that never occurred.
The scenarios look worlds apart, yet they converge on the same unsettling question: what happens when citizens can no longer tell the difference between legitimate authority and unchecked power?
This is the exact alarm Justice Sonia Sotomayor raised in her dissent in Trump v. United States (2024). By expanding presidential immunity, the Court, she argued, had crossed a dangerous threshold: “In every use of official power, the president is now a king above the law.”
Her words strike at a fracture point in democratic life: the erosion of accountability. If citizens lose the ability to distinguish between a president and a king, the foundation of checks, balances, and the rule of law begins to crack.
From my own work designing philosophical architectures for AI, I cannot help but hear Sotomayor’s warning resonate in another domain. In government, an unchecked president risks becoming a king. In technology, an unaccountable AI risks becoming an oracle, one capable of reshaping reality itself and turning information into a weapon of perception.
The danger is convergence. If citizens lose both constitutional literacy and informational trust, democracy may not collapse under force but dissolve quietly into rule by perception rather than rule by law.
Constitutional Design: A President, Not a King
The framers of the Constitution confronted a paradox. How could they construct an executive strong enough to govern yet limited enough to avoid monarchy? Their solution was architectural rather than personal.
Congress would make laws. The President would enforce them. The Courts would interpret them.
Alexander Hamilton, writing in Federalist No. 69, drove this distinction home. Unlike the British monarch, he insisted, the American president was “an elective magistrate of four years.” Temporary. Limited. Accountable.
This was never about trusting the goodwill of leaders. It was about embedding accountability directly into the design of governance. The Constitution functioned as a trust infrastructure, a set of systemic safeguards built not to assume virtue but to contain vice.
Case Law: Stress Tests of the System
Yet even the strongest architecture must be tested by history. And history has pressed hard against these constitutional guardrails.
Youngstown Sheet & Tube Co. v. Sawyer (1952): Truman could not seize the steel mills without Congress.
United States v. Nixon (1974): Nixon was compelled to release his tapes; no president stands above the law.
Trump v. United States (2024): Presidential immunity was expanded, prompting Sotomayor’s dissent that the presidency had drifted into kingship.
Each case shows the Constitution functioning as a living stress test. At times the guardrails hold. At other times they bend. Sotomayor’s dissent raises the question of whether those guardrails can still restrain the pull toward monarchy in practice.
Civic Literacy: The First Line of Defense
Sotomayor’s warning is not simply about constitutional architecture. It is about citizens themselves. No set of parchment barriers can defend democracy on their own.
If citizens cannot distinguish lawful authority from arbitrary power, they cannot resist overreach. Which means civic literacy is not a luxury. It is the first line of defense.
What would this look like in practice?
Reforming K–12 education to place civics alongside math and science.
Expanding adult education to illuminate how government actually functions.
Promoting media literacy to help citizens spot propaganda and misinformation.
Constitutions do not enforce themselves. They require citizens who can recognize kingship even when it dresses in robes of legitimacy.
The Parallel with AI
Unchecked presidential power is not the only path to kingship. Unchecked AI introduces a parallel danger. Both collapse trust. Both blur the line between authority and truth.
Without limits, AI can generate misinformation at scale. Without transparency, citizens cannot verify what they see. Without pluralism, narrative control concentrates in the hands of a few.
The pattern is familiar: erosion of accountability, consolidation of power, the substitution of law with perception.
If democracy requires limits on presidents, then the information ecosystem requires limits on AI. Sotomayor’s dissent, while aimed at the presidency, applies here as well: unchecked authority, whether political or technological, drifts toward monarchy.
Constitutional Principles as a Blueprint for AI Governance
The symmetry between constitutional design and AI governance is striking. What kept monarchy at bay may also prevent AI from becoming an informational monarchy.
Consider checks and balances. In democracy, this principle prevents any branch from consolidating power. In the information ecosystem, the equivalent is auditability and oversight: independent review processes that prevent narrative capture and the quiet consolidation of informational control.
Or take the rule of law. No one is above it. In AI, this finds a parallel in traceable accountability: every output should link back to responsible human actors. Otherwise, we drift into a space where decisions carry authority but no one can be held accountable for them.
Separation of powers provides another echo. In constitutional design, the functions of government are divided among branches. In AI, governance can be modular, dividing responsibility across data collection, model training, and deployment rather than concentrating it in one opaque entity.
Even the principle of limited terms has resonance. Presidents cannot rule indefinitely. AI systems, too, should not remain static and unexamined; lifecycle controls—versioning, updates, and decommissioning—ensure that models do not linger past their ethical or technical relevance.
Judicial review provides a further parallel. Courts check overreach, ensuring that constitutional limits remain binding. In AI, we need algorithmic oversight: independent audits, appeals, and mechanisms for redress when systems produce harm.
Finally, civic literacy, the citizen’s ability to recognize illegitimate power, finds its counterpart in AI literacy. Users must be able to detect bias, recognize manipulation, and resist mistaking machine outputs for truth. Without such literacy, the architecture itself cannot hold.
The lesson is simple yet profound. The very structures that restrained monarchy in government may hold the key to preventing AI from becoming an unelected ruler in the domain of information.
Closing Reflections: Sotomayor’s Question for the Information Age
Justice Sotomayor warned that constitutional illiteracy risks turning presidents into kings. I agree with her—and I believe her warning applies with equal force to artificial intelligence. The greatest danger is not that AI grows powerful, but that citizens remain illiterate about it.
Two blind spots matter most:
Illiteracy about AI’s limits. Too many mistake outputs as neutral or objective, when in fact they are products of design choices and embedded biases.
Illiteracy about AI’s accountability. Too often we fail to ask who bears responsibility for what these systems generate.
If we leave these gaps unaddressed, we may wake one day to find that the line between tool and ruler has already blurred. By then, AI may occupy the same place monarchy once did: unelected, unchallenged, and deeply embedded in the structures of power.
At that point, restoring accountability would be nearly impossible, because the very capacity to distinguish truth from authority would have eroded.
Sotomayor asked whether Americans can still tell the difference between a president and a king. The corresponding question for our time is whether we can still tell the difference between a tool and a ruler. Our answers will determine not only whether democracy survives in law, but whether truth itself survives in the information age.
Works Cited
Hamilton, Alexander. Federalist No. 69. 1788.
Youngstown Sheet & Tube Co. v. Sawyer. 343 U.S. 579. Supreme Court of the United States. 1952.
United States v. Nixon. 418 U.S. 683. Supreme Court of the United States. 1974.
Trump v. United States. Supreme Court of the United States. 2024.
Congressional Research Service. Presidential Immunity from Criminal Prosecution in Trump v. United States. 2024.
Harvard Law Review Forum. “Disqualification, Immunity, and the Presidency.” 2025.
Key Concepts and Working Terms
Trust Infrastructure: Systemic safeguards built into governance or technology to keep power accountable.
Informational Monarchy: A condition where one actor or system controls narratives without transparency or oversight.
Lifecycle Controls: Governance mechanisms for versioning, expiration, and decommissioning of AI systems.
Civic Literacy: Public knowledge of constitutional structures and limits.
AI Literacy: Public understanding of AI’s limits, biases, and accountability.
