For years, warnings about the risks of artificial intelligence came from outside. Scientists, philosophers, governments, and even the Church questioned how far to go and what the consequences might be.

Now the warning comes from a more disturbing place:  from those who build it .

Dario Amodei, CEO of Anthropic, has just called on major laboratories to slow the pace of development of their most powerful AI models. He’s not proposing to halt artificial intelligence development, but rather to avoid advancing faster than we are capable of understanding and controlling it.

Sam Altman, CEO of OpenAI, endorsed the need to “set the pace” of the technological frontier and supported independent evaluations of the models. Elon Musk was even more succinct: “Dario is right.”

The question then seems inevitable:

Who’s up for taking their foot off the gas?

The answer is not only technological. It is also political and economic.

China acknowledges the risks. It is developing security standards and mechanisms to keep AI systems under human control. But it rejects the notion that security concerns should be used to limit its technological development.

Beijing warned that “alarmism, confrontation and fierce competition” harm the global governance of artificial intelligence.

Donald Trump chose a very different response. He downplayed the warnings about the risks of AI and summarized his position in one sentence:

“Whoever wins AI, wins.”  Whoever wins the artificial intelligence battle, wins.

That’s where the dilemma arises.

Everyone can acknowledge that risks exist. But who is willing to slow down when doing so could mean someone else gets there first?

Months before this discussion, Leo XIV had proposed another way of looking at the problem.

In  Magnifica Humanitas,  he argues that calling for “prudence, rigorous controls” and even, in certain circumstances,  slowing down the adoption of artificial intelligence does not mean being against progress .

The Pope warns of an increasingly evident imbalance: technology is advancing at a much faster pace than our ability to understand its consequences and build the norms, institutions and consensus necessary to govern it.

Technology learned to run faster than we did.

That is why Leo XIV sets the limit elsewhere: artificial intelligence must remain under human responsibility. Innovation cannot end up deciding for itself where we are going.

And it also raises an uncomfortable question: if we want AI to respond to certain “human values”,  who decides what those values ​​are?

The companies that develop the models, the governments, the United States, China, those who own the data or those who control the technology.

Because whoever defines the values ​​of an artificial intelligence also acquires enormous power to define the world in which that intelligence will act.

Perhaps that’s why the discussion can no longer be reduced to being for or against artificial intelligence.

Amodei calls for slowing down. Altman agrees that the pace needs to be regulated. China acknowledges the risks but doesn’t want to lose ground. Trump responds that whoever wins the AI ​​battle wins.

And Leo XIV introduces a different question.

Not who gets there first, but how we want to get there.

We can build increasingly faster, more capable, and more autonomous machines. We can make them do tomorrow what still seems impossible today.

But an artificial intelligence doesn’t feel joy or pain. It doesn’t love. It has no friends. It doesn’t know fragility, hope, or fear. It can process millions of texts that describe these experiences, but it can’t live them. We can.

Therefore, perhaps the greatest challenge in this race isn’t making artificial intelligence increasingly like us. Rather, it’s that in our obsession with being first,  we end up becoming more and more like machines , leaving our humanity behind.