Pope Leo XIV's Encyclical Nails the Real AI Governance Problem

On May 15, Pope Leo XIV signed an encyclical warning that AI’s concentration of power amounts to ‘algocracy’—algorithmic governance with no democratic mandate, no regulatory framework, and a handful of private actors deciding the values encoded into systems used by billions.

The Structural Problem Self-Regulation Can’t Fix

Magnifica Humanitas flags autonomous weapons, destabilized democratic discourse, and predictive algorithms already informing policing strategies and sentencing decisions. The Pope’s diagnosis: the prevailing approach in the US—voluntary ethics commitments rather than enforceable regulations—is not governance, it’s abdication.

I’ve sat in policy rooms with AI developers and government ministers. The knowledge gap isn’t technical. Engineers understand their tools in extraordinary detail. What they resist seeing is that optimizing a content-recommendation algorithm for clicks is a political act, not an engineering decision. Choosing to maximize engagement often means favoring outrage. That’s a value choice, not a benchmark metric.

Why a Religious Text Matters to Technical Audiences

The Catholic Church invokes its full teaching authority on social questions only when secular institutions—universities, parliaments, scientific academies—have been unable or unwilling to address a crack in human coexistence. In 1891, Pope Leo XIII warned that industrial automation could reduce workers to interchangeable commodities. In the 1930s, encyclicals responded to the rise of totalitarianism, Nazism, and Stalinism. Magnifica Humanitas rings the same civilizational alarm bell.

The author, an advisor to the UN and the Vatican’s RenAIssance Foundation, notes that scientists who maintain studied neutrality are complicit in the status quo. When those profiting from AI also write its ethics frameworks, the result isn’t accountability—it’s a PR exercise.

What Real Governance Looks Like

Taking the encyclical seriously means AI researchers treating governance as a professional responsibility, not a side project. Mandatory algorithmic impact assessments—modeled after environmental impact assessments for power plants—should be standard. Third-party auditing of systems in criminal justice, credit, and health care should be deployed. And the communities whose labor and information systems are being restructured need a seat at the standards-setting table, including the UN and technical bodies.

Leo XIII didn’t stop industrial capitalism. The 1930s encyclicals didn’t prevent totalitarianism. But they named what was happening and provided a framework for moral judgment. Magnifica Humanitas does the same for AI. Ignoring it won’t make the crack smaller.

Source: I advise the Vatican and the UN on AI - don't dismiss the Pope's message as theology
Domain: nature.com