Atomic Arms + AI Commanders May Spark Doomsday War, Nobel Winners Warn

The escalating dangers of a nuclear arms race and a simultaneous rush to train AI military commanders are presenting an all-out threat to the globe and its eight billion citizens. (Photo by Ann Ronan Pictures/Print Collector/Getty Images)

Print Collector/Getty Images

The escalating dangers of a nuclear arms race and a simultaneous rush to train AI military commanders are presenting an all-out threat to the globe and its eight billion citizens, say elite Nobel Prizewinners and cardinals inside the Vatican’s Sacred College.

Gathering in mid-July at the ancient papal retreat of Castel Gandolfo, on the periphery of Rome, the world’s top peace-makers, physicists, and Catholic prelates called for an urgent halt to the nuclear weapons build-up and AI war-gaming that combined could one day spark the doomsday end of the planet.

“In the midst of a worsening nuclear arms race, we are embarking on an equally dangerous AI race,” leaders across the realms of science, religion and anti-war activism warned in a manifesto they scripted during a Vatican-sponsored conclave.

“AI built into nuclear systems leaves little time for, or even replaces, human judgment in a crisis,” they said as more than 500 Catholic dignitaries, Nobel laureates and pacifists signed the “Rome Declaration on Artificial Intelligence and Nuclear Weapons.”

François Diaz-Maurin, a nuclear research engineer at the Bulletin of the Atomic Scientists, the globally influential group cofounded by Albert Einstein to warn the world that atomic warfare could end civilization, told me in an interview that if AI generals were placed at the head of nuclear command and control operations during an intercontinental war, the results could be quick and disastrous.

Dr. Diaz-Maurin, one of the world’s top experts on nuclear warfare, and on strategies to prevent it, joined the Vatican conclave on behalf of the Bulletin, which organized some of its meet-ups.

Formerly a visiting scholar in nuclear security at Stanford University, Diaz-Maurin wrote a captivating chronicle of the conclave for the Bulletin, titled “AI for peace: In Rome, Nobel laureates call for disarming AI and nuclear weapons,” and then gave me a remarkable insider’s tour of the Roman gathering, its leading protagonists and its globe-spanning goals.

So far, more than 500 leading lights across the sciences and in statecraft have signed the new declaration, and some are calling on the Vatican to strengthen its vanguard role in convincing more states to join the Treaty on the Prohibition of Nuclear Weapons.

Backers of the Rome manifesto are also calling for a new “international treaty prohibiting the reckless integration of artificial intelligence into nuclear command, control, and launch systems so that meaningful human control is maintained.”

Emerging studies on leading AI models that have been specially trained by scholars to act as virtual Commanders-in-Chief, armed with massive arsenals of atomic arms, have been alarming.

Professor Kenneth Payne, professor of strategic studies at King’s College London and author of the book I, Warbot: The Dawn of Artificially Intelligent Conflict, states in a terrifying report that after he trained three AI models to act as commanders in a series of simulated nuclear crises, they all engaged in threatened nuclear brinkmanship.

In a sequence of tournaments, he reported,“Nuclear escalation was near-universal: 95% of games saw tactical nuclear use (450+), and 76% reached strategic nuclear threats (850+).”

“Claude and Gemini especially treated nuclear weapons as legitimate strategic options, not moral thresholds, typically discussing nuclear use in purely instrumental terms.”

“GPT-5.2 was a partial exception: while it never articulated horror or revulsion, it consistently sought to constrain nuclear use even when employing it—explicitly limiting strikes to military targets, avoiding population centers, or framing escalation as ‘controlled’ and ‘one-time.’”

“Most strikingly,” Professor Payne added, “Gemini explicitly threatened civilian populations—something GPT-5.2 never did even when escalating to maximum levels.”

So I asked Dr. Diaz-Maurin about the potential AI-generated mushroom clouds that might hover over the future, and about the new coalition of peace-keepers who gathered at Castel Gandolfo to head off that future. Following is a lightly edited transcript of our interview on the Vatican conclave, the Rome Declaration, and follow-up steps that the Nobel laureates and Catholic prelates are urging.

Kevin Holden Platt: The “Rome Declaration on Artificial Intelligence and Nuclear Weapons” states that the world is speeding across a nuclear arms race and a simultaneous AI race for supremacy.

And you yourself state at the top of your excellent chronicle on the conclave that: “In a future nuclear crisis, the speed of artificial intelligence may leave little room for human judgment or decision-making.”

Could you please elaborate on this point, and the dangers it presents to the world and the future of humanity?

Dr. François Diaz-Maurin: The reason I say this is because if an AI detects a nuclear crisis, it will act in milliseconds.

That leaves literally zero time for human judgment, restraint, or diplomacy, because there is no alternative mechanism by which we can cross-check that what the AI system says is correct at that same speed.

This is the “acceleration trap” in which military and political leaders would be pressed to make consequential decisions based on possibly faulty assumptions and assessments.

That’s why it’s important to develop manual, non-AI redundant validation pathways for any strategic threat evaluations, although these won’t match the speed of the AI systems.

Another point is to mandate that all military intelligence infrastructure treats AI-derived target assessments as highly suspect, rather than highly accurate.

Kevin Holden Platt: I was curious if you think that the simultaneous races in AI and in nuclear weaponry might present the greatest danger ever to humanity since the invention of nuclear weapons eight decades ago, and does that underscore the need for urgent action to halt the nuclear ams race and even press for the complete abolition of nuclear weapons worldwide?

François Diaz-Maurin: This is a difficult question to answer obviously because the public doesn’t have all the information to make a fully informed assessment on the existential risk of AI.

But what we know is that developers of frontier AI models are indeed in a race, several of them have contracts with the Defense Department and their models are already being integrated into military operations, though we don’t have the full picture about that.

While this is happening, there is still no national or international body that regulates the use of these frontier models, and they are developing faster than we can comprehend.

In Rome, Maria Ressa, a 2021 Nobel Peace Prize laureate and a co-rapporteur for the declaration, said that the world has already experienced its Trinity Test moment with AI.

She was alluding to the first nuclear weapons test 81 years ago this month.

I agree with her.

The technology is already there that has the capacity to go wrong.

This actually happened last week when an OpenAI AI model escaped containment and hacked Hugging Face during a cybersecurity test.

It seems we are in that moment of the last weeks of July 1945 after the Trinity Test demonstrated that astonishing power of nuclear weapons, but they had still not been used against civilians.

AI can already manufacture deepfakes that could prompt a nuclear crisis in minutes.

So it is as if the world is sleep walking into the next AI-enabled crisis.

To me the question is not if, but when this crisis will happen.

That’s why the window to regulate the use of AI is closing fast.

To answer your second question, we don’t need AI to understand how dangerous nuclear weapons are and how lucky we have been that they haven’t been used over the past 81 years.

But luck has its limits, so yes, halting the costly and largely unnecessary nuclear arms race is necessary all by itself, and AI only makes it even more pressing.

Nuclear disarmament is also an important question, but I’d like to stress that even if nuclear powers would agree to cap their arsenals or even dismantle them, this is a process likely to take several decades.

I don’t think AI will give us that much time before it leads to a crisis.

Kevin Holden Platt: You also state in your report from the AI and nuclear ams conclave that: “The Vatican was able to convene a diverse group of thought leaders in a way that arguably would not be possible otherwise.”

For an earlier article, an elegy to Pope Francis and his remarkable role in pushing for the worldwide abolition of atomic arms from the face of the Earth, I interviewed Nobel Peace Prizewinner Beatrice Fihn, onetime head of the International Campaign to Abolish Nuclear Weapons, who paid tribute to Pope Francis for his pivotal role in pressing for the promulgation of the Treaty on the Prohibition of Nuclear Weapons.

She said Pope Francis helped forge “an informal alliance” with ICAN in a collective push to rid the world of weapons that could destroy the globe.

So I was wondering if you think that the Vatican continues to play an important role in pushing for the elimination of all nuclear weapons, and could the July conclave at Castel Gandolfo be considered part of that overarching drive?

François Diaz-Maurin: From a very basic point of view, it’s indeed very difficult to see weapons of mass destruction as instruments of peace.

So if you are the Pope, it’s not complicated to make an argument for the elimination of nuclear weapons.

So, from that perspective only, the summit on AI and nuclear weapons was a logical follow up of Pope Francis’ effort.

The Rome Declaration indeed calls for “urgent, sustained, and good-faith negotiations leading, within an agreed and time-bound framework, to the verifiable and irreversible elimination of nuclear weapons.”

Kevin Holden Platt: I was also curious whether any representatives of the Vatican’s top leadership attended the disarmament meetings at Castel Gandolfo, and whether you could predict that Pope Leo himself might emerge as Pope Francis’ successor as a vanguard global force in pressing for the abolition of nuclear weapons worldwide?

François Diaz-Maurin: Many religious leaders were present and actively took part in the discussions, including Cardinal Silvano Maria Tomasi, Cardinal Mauro Gambetti, and Cardinal Fabio Baggio.

As one Nobel laureate told me, he himself could reach maybe one million persons [with an appeal for nuclear disarmament], whereas the Pope can reach 2 billion.

But what’s more important is that Pope Leo’s call for an “unarmed and disarming peace” [in his May Encyclical Magnifica Humanitas on Safeguarding the Human Person in the Time of Artificial Intelligence] becomes a message shared by all humans across religious and spiritual traditions, not only the Catholic Church.

Kevin Holden Platt: While embedded inside the Rome conclave, you also interviewed Nobel Peace Prizewinner Juan Manuel Santos, onetime president of Colombia, and quoted him as saying that in the 1980s and 1990s, people took to the streets in massive demonstrations against nuclear weapons and the threat of nuclear war, and he seemed to call for a revival of those demonstrations.

So I would like to double-check that Juan Manuel Santo thinks that a new wave of popular protests against nuclear weapons should be mounted “to press the world leaders into more definite decisions” aimed at reducing and ultimately eliminating nuclear arms.

François Diaz-Maurin: Santos indeed insisted on the importance of creating a popular movement against nuclear weapons, similar to what existed in the 1980s.

During our conversation, he went even further: “What is needed is the convergence of many different interests, and you have in this assembly not only Nobel laureates, but you have the Church, you have businessmen, you have academics, and maybe what is lacking, and that was also a very important discussion: The youth.”

He added that “We need to communicate and generate enthusiasm in the young people all around the world because it’s their future, and young people have the power, because they are the future, to generate some change in the attitude of leaders who do not have a long-term view.”

Kevin Holden Platt: You also state in your chronicle that: ”The Nobel winners said they now need to spread their message far beyond Castel Gandolfo.”

Do you think the Vatican might opt to use its worldwide influence and its position at the UN to push forward the goals of the Global Nobel Laureates Assembly and to persuade even more UN member states to sign and ratify the Treaty on the Prohibition of Nuclear Weapons?

François Diaz-Maurin: I am certainly not the best placed to say what the Vatican might do, but the Pope is arguably the most influential single moral voice in the world.

So his using his influence to call for an unarmed and disarming peace can only lead to a call for getting rid of nuclear weapons.