Pope Leo XIV's first encyclical, released in May 2026, raises an alarm that goes beyond theology: a handful of private actors are making decisions about artificial intelligence that affect billions of people, with no democratic mandate and minimal regulatory oversight. The document, titled "Magnifica humanitas" (magnificent humanity), diagnoses a governance crisis that theologian Paolo Benanti argues the scientific community has been too quiet about.

Why Is a Religious Document Suddenly Central to AI Policy Debates?

The encyclical's focus on AI governance might seem surprising coming from the Vatican, but Benanti explains that the Catholic Church historically invokes its full teaching authority when secular institutions fail to address fundamental threats to human society. During the Industrial Revolution, Pope Leo XIII warned in 1891 about automation reducing workers to interchangeable commodities. In the 1930s, papal encyclicals responded to the rise of totalitarianism. "Magnifica humanitas" is ringing the same civilizational alarm bell, according to Benanti.

The document flags several interconnected governance problems that extend far beyond religious concerns:

• Power Concentration: A small number of private companies control how AI systems reason, what they optimize for, and whose values they embed, without democratic input or accountability.
• Algorithmic Governance: Systems called "algocracy" allow algorithms to make decisions previously made by humans, including in policing strategies and criminal sentencing, with minimal oversight.
• Regulatory Gaps: AI development operates largely beyond national regulatory frameworks, relying instead on voluntary self-regulation and ethics commitments rather than enforceable rules.
• Democratic Deficit: Billions of people are affected by AI systems designed by private actors without their consent or knowledge.

Benanti argues that some scientists maintain studied neutrality on these issues, effectively becoming complicit in the abdication of governance responsibility.

Where Are the Real Gaps in AI Oversight Today?

The governance crisis extends beyond philosophical concerns into concrete regulatory failures. In healthcare, a sector where AI decisions directly affect patient outcomes, the oversight landscape is fragmented and incomplete. Maya Sandalow, associate director for the Bipartisan Policy Center's health program, explains that health AI tools fall into two broad categories: administrative AI (used for scheduling, benefits navigation, and claims processing) and clinical AI (used to diagnose or treat disease).

The problem is that the same AI tool may be regulated by different federal agencies depending on where it is used and what data it processes. The Food and Drug Administration (FDA) oversees medical devices, but many health AI applications don't qualify as medical devices and thus escape FDA review. Meanwhile, the Centers for Medicare and Medicaid Services (CMS) handles coverage decisions, the Office of Civil Rights enforces patient privacy rules under the Health Insurance Portability and Accountability Act (HIPAA), and the Federal Trade Commission (FTC) can intervene if marketing claims are misleading.

"The same AI may be regulated by different entities depending on where it is used, the data that it's trained on," explained Maya Sandalow.

Maya Sandalow, Associate Director for the Bipartisan Policy Center's Health Program

This fragmentation creates significant risks. Administrative AI tools like ambient scribes (which record and transcribe patient conversations) and prior authorization systems (which determine insurance coverage) are among the fastest-growing AI applications in healthcare, yet they often fall outside traditional medical device oversight. The regulatory framework is so complicated and payment mechanisms so unclear that there's a "chilling effect" on deploying clinical AI tools that could genuinely help diagnose disease or prevent illness.

How Can Organizations Navigate AI Governance in Practice?

For organizations trying to adopt AI responsibly, the challenge is that governance frameworks are still being built. At Georgetown Law, researchers working through the Georgetown AI Law & Policy (GAILP) program are developing practical tools to help legal practitioners and organizations implement AI thoughtfully. Chance Goddard, a recent graduate who studied both artificial intelligence and law, is part of a team creating what they call a "Maturity Model" for AI implementation.

• Stage-Based Implementation: The model identifies approximately five stages of AI adoption in organizations, with the goal of supporting practitioners at each stage to help them personally adopt AI and become champions of responsible AI use.
• AI Fluency Development: Rather than treating AI as a tool to deploy and forget, the framework emphasizes helping legal professionals develop genuine AI fluency so they can use these systems in ways that reflect ethical understanding and professional responsibility.
• Bridging Research and Practice: The approach combines technical knowledge about how AI works with legal and policy expertise, recognizing that effective governance requires both understanding the technology and understanding the regulatory landscape.

"We are tracking these changes to understand how AI is affecting the industry, and we are future-casting based research to highlight gaps," noted Chance Goddard.

Chance Goddard, Graduate, Georgetown Law Tech Law Scholars Program

Goddard's experience illustrates why bridging technical and legal expertise matters. Before law school, he earned a doctorate in information technology focused on AI in educational settings, which led him to recognize that the issues he was researching had significant legal components. "I was unsure whether I could be effective in that space without actual legal knowledge and the ability to advocate within the legal space," he explained. His dual background now positions him to help organizations understand not just how AI works, but how to implement it in ways that align with ethics and professional responsibility.

What Happens When Governance Lags Behind Deployment?

The core problem identified by both the papal encyclical and policy researchers is that AI is moving faster than governance frameworks can accommodate. In healthcare, the most common AI applications today are administrative rather than clinical, partly because the regulatory pathway for clinical AI is so unclear. This means the tools with the greatest potential to improve diagnosis and treatment are being held back by uncertainty, while less-scrutinized administrative tools proliferate.

Benanti argues that deferring to self-regulation, as the United States primarily does, is "more an abdication of responsibility than governance." The Vatican's intervention suggests that when democratic institutions fail to establish clear rules, other institutions may feel compelled to speak up. The question now is whether policymakers will respond to these warnings by building more coherent, enforceable governance frameworks before AI systems become even more entrenched in critical decisions about healthcare, criminal justice, and democratic discourse.

Benanti