At the University of Arizona’s commencement ceremony last month, former Google CEO Eric Schmidt told the class of 2026 that they would shape AI’s future. The crowd booed him. Other commencements followed with a similar pattern: a hopeful line about AI, then rejection from the crowd.
Their anger is real, and it is justified. Tech executives have told us that AI will remake our jobs, schools, relationships, and wars. But we were also asked to believe we would all benefit from a change unfolding without our consent or participation. For many Americans such as myself, that promise doesn’t fit: we sense that our future is being decided for us, rather than by us.
The past several decades of technology policy have been treated as a narrow technical affair, the domain of scientists, industry CEOs, procurers, and lawyers. This spring, a Pope and a President made it a moral and political question. On May 25, Pope Leo XIV published “Magnifica Humanitas,” his first encyclical on AI and the human person. Then, on June 2, President Donald Trump signed an Executive Order titled “Promoting Advanced Artificial Intelligence Innovation and Security.” The two documents could hardly differ more.
The Pope’s encyclical focuses on humanity and demands that power be constrained by morality. On the other hand, Trump’s order asks how America stays ahead of China. The first is a meditation on dignity, the second a strategy for dominance. Yet neither statement is only about AI. Both turn on a shift: authority passing from citizens to machines, and from public debate to private hands.
The gap between the two documents is widest on the topic of war and national security. The encyclical’s strongest warning is that lethal and irreversible decisions must never be entrusted to artificial systems. A machine should not decide who lives and who dies. A weapon can read a heat signature, track a figure, and recommend a strike, but cannot understand what it means to end a life. It cannot answer to a family, a court, or history. In the words of Pope Leo, “No algorithm can make war morally acceptable.”
Trump, for his part, has entered the U.S. into multiple conflicts abroad, including an ongoing war with Iran. As such, his order approaches frontier systems from a more tactical angle and establishes that Trump Administration officials will test the most advanced models in secret. It does not explicitly authorize autonomous weapons, but it defines where decisions regarding frontier AI systems and their use in war will be made: behind closed doors. Going forward, the gravest questions about AI’s military, cyber, and social power will be answered through a classified review and private collaboration between figures such as the Secretary of the Treasury, Scott Bessent; the Secretary of War, Pete Hegseth; and the Secretary of Commerce, Howard Lutnick.
A democracy should settle questions like these in the open, with a hearing, a vote, or a record that someone could be held to. Now, this order could move them further out of public reach, where few citizens have standing to be heard.
More broadly, Trump’s EO hands almost the entire review process to the government and a handful of organizations named as “AI developers and researchers”, or as the “AI industry”. Going forward, the agencies will determine which models should be released to market through a classified benchmarking process, in collaboration with the companies that designed them. Together, they choose the “trusted partners” who see the most advanced systems before the public does, and the government takes its own look at them up to 30 days before a model even reaches those partners. The order opens no comment period, schedules no hearing, and creates, in its own words, no right “substantive or procedural” that anyone can enforce against it. Participation is voluntary, which is to say it runs by invitation, and the developers who help design the framework help choose who else gets in. The public reads about it afterward.
Even before the order was signed, this dynamic was on full display. Axios reported that Mark Zuckerberg, Elon Musk, and David Sacks spoke with Trump before an earlier draft was postponed, while The Washington Post described industry calls opposing parts of the language. Whatever happened inside those conversations, they revealed a deeper imbalance: the companies building these systems already have a direct channel into the rules that will govern them, while the public is still waiting for a way in.
That brings us to the Pope’s encyclical, which gives language to the danger underneath such a bargain. When data, algorithms, platforms, and infrastructure are concentrated in a few hands, goods that should widen human possibilities can become instruments of exclusion. Will intelligence become shared infrastructure, open enough for public institutions and communities to shape? Or will it become rented capacity, sold back to the public by a few companies with the capital to build it?
To be sure, this evolution is already underway. AI already impacts people directly: in what a machine decides about them, and in what it may take from their lives. Software already ranks a child’s academic promise, prices a loan, sorts who receives care, and decides which job applications a human ever sees. As these systems spread, and more of our ordinary lives runs through them, judgment will increasingly arrive with no way to see how it was reached.
Here, Leo’s encyclical’s defense of human dignity is key. The parent, the patient, the worker, the student, the job applicant: each can be bound by a verdict from a system they did not choose and cannot appeal. In a courtroom, any of them would have standing. In a life-altering automated decision, they may have none.
The fear that any pause to assess how AI impacts people will hand the future to China runs through Washington and Silicon Valley alike: move faster, ask less, trust the builders. But a free society cannot treat speed as a substitute for consent. A shift like this happens slowly, and then all at once: in procurement contracts, model-access agreements, and clauses no one reads, until the systems are built, the defaults set, and the arrangement too expensive to undo. We are still in the slow part.
The solution is ordinary democracy: a public hearing before a school district buys a grading algorithm, the right to know when a model has judged you and to appeal, public review of frontier-model decisions outside the classified state, and procurement rules that ask who benefits, who is exposed, and who can challenge the outcome.
I return to a powerful invitation in Magnifica Humanitas: “Let us not be afraid to get our hands dirty on the ‘construction site’ of our time.” The students who booed this spring already know what took policymakers years to admit: the construction site has been walled off. The decisions are being made inside. A free people can still own them, but only by refusing to quietly wait to be told what was already decided.
The post The Fight Over AI Is Really a Fight Over Who Governs appeared first on TIME.
