Magnifica Humanitas and AI: What the Newspapers Got Wrong

"The Pope wants to disarm AI", almost everyone headlined. But that's exactly the point the newspapers missed.

When the Vatican published Magnifica Humanitas, the first encyclical of Pope Leo XIV, nearly all the newspaper headlines said more or less the same thing: "the Pope wants to disarm AI". A simplification that, according to many observers in the tech sector, completely missed the point.

One of them, Raffaele Gaito, an entrepreneur, investor and one of Italy's most followed digital commentators, devoted a video to the document, calling it "incredible", "historic" and "extremely on point". An unexpected, and deliberately secular, judgment.

It's worth understanding why.

Raffaele Gaito's video

Gaito analyzes Magnifica Humanitas with a practical, non-confessional approach, focusing on the implications for business and technology.

The six points Gaito highlights

1. The newspapers didn't read it

Gaito is blunt: most media outlets wrote headlines without reading. The encyclical isn't a document against technology. On the contrary, it's pro-technology and pro-progress, with one condition: that progress be guided by discernment, not by blind optimization.

2. The problem isn't AI. It's who controls it.

One of the document's densest passages concerns the concentration of power in the hands of a few transnational companies, those of Silicon Valley, which today have resources greater than many states and govern innovation without any democratic mandate.

The encyclical directly targets the narrative of certain tech CEOs who promise to "fix" humanity or free it from fragility through technology. Gaito calls these promises what they are: deceptive.

3. You don't delegate what requires compassion

The most concrete point, and perhaps the most important for anyone working in business: you cannot delegate to machines the decisions that affect people's rights, reputation and freedom.

Hiring, mortgages, medical priorities, access to services. An algorithm has no compassion, no political responsibility, cannot be held to account. Handing it decision-making power means hiding the "discarding of the weak" behind a false objective neutrality. Responsibility doesn't disappear, it just gets masked.

4. Algorithms, data and patents as common goods

Gaito shares the document's thesis on the need to treat algorithms, data and patents as goods universally destined for everyone, not as private assets to be monetized without limits. The concrete risk, already visible, is a new divide between the included and the excluded: no longer based on income, but on access to cognitive infrastructure.

5. "Disarming AI" doesn't mean dismantling it

This is the most misunderstood point, and the one Gaito insists on most. Disarming AI doesn't mean stopping technological development. It means three precise things:

• removing it from the logic of armed competition (military, economic, cognitive);
• breaking the equivalence between technical power and the right to govern;
• making it debatable, contestable and therefore livable, preventing it from dominating instead of serving.

6. Ethics isn't enough if companies decide it

Gaito appreciates the document's pragmatism on a point that anyone in the sector knows well: it isn't enough to invoke ethics or "alignment" if these values are defined autonomously by the very companies that build the models. What's needed is an active politics, capable of slowing processes down enough to let communities take part in the choices.

Media Engineering's point of view

We've been building AI systems for companies for years. And some of the questions the encyclical raises, who decides, who answers, who gets excluded, we hear every week in conversations with our clients.

Not from a religious perspective. From a practical one.

• Who's accountable when AI gets it wrong? When an automated system denies access, suggests a priority, selects a candidate, someone has to answer for it. The point isn't whether AI can do it more efficiently than a human. The point is that responsibility can't evaporate into the model.
• Data sovereignty isn't optional. One of the most debated topics at AI Week 2026, and one we work on actively, is exactly this: building AI systems that run on the client's infrastructure, without data leaving the company perimeter. It isn't just a compliance issue. It's a question of who owns the knowledge that's generated.
• AI must be contestable. This is the sentence from the encyclical that resonates most with our daily work. Building explainable, auditable, modifiable systems isn't a cost: it's a precondition for using them in high-responsibility contexts such as healthcare, training, recruiting and credit.

We don't agree with everything the document says, and we don't need to. But on this, that AI should be a tool at the service of people and not an autonomous entity that decides in their place, there isn't much to debate.