One year after issuing an historic statement on the risks of the new nuclear arms race, a group of Nobel Laureates gathered outside of Rome to address the looming challenge of combining nuclear weapons and artificial intelligence. The Global Nobel Laureates Conference on Nuclear Weapons and Artificial Intelligence met for two days at Castel Gandolfo, the historic summer residence of the Pope outside of Rome, and one day later issued the Rome Declaration: Humanity at the Threshold: A Declaration on Artificial Intelligence and Nuclear Weapons on July 16, 2026.
The two-day meeting included over 30 Nobel Laureates, experts in nuclear weapons, artificial intelligence, diplomacy, finance and other fields, and included attendees from over 30 countries. The event itself was sponsored by the Nobel Laureates as well as the Vatican, the University of Chicago, the Bulletin of the Atomic Scientists and a variety of other global non-governmental organizations.
PAX sapiens US Nuclear Policy Fellow Jon Wolfsthal attended the conference and spoke on the initial panel along with Nobel Prize Winners Maria Ressa and Brian Schmidt of Australia National University. Mr. Wolfsthal also published an OpEd in the New Republic and spoke with PBS NewsHour on the topic.
In the end, the Rome Declaration made clear the deep and unaddressed concerns regarding the integration of artificial intelligence into nuclear weapons command and control, while noting the potential benefits AI had to offer the world. The Declaration leads by noting that “Never before has scientific progress offered such extraordinary opportunities while simultaneously creating such profound risks to our common future.” In addition to calling for increased international cooperation and engagement to secure the benefits of AI for all of humanity, the statement also calls for an international agreement to ensure that an automated system should never make the final decision to launch a nuclear weapon. In addition to restating the 2025 call for states with nuclear weapons to fulfill their obligations to negotiate an end to the arms race and pursue nuclear disarmament, the statement also calls for responsible international engagement and governance over AI to enhance its benefits and disarm its potential risks.
Both the conference and the statement were heavily influenced by the recently issued Encyclical from Pope Leo XIV Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence. Mr. Wolfsthal noted that “human fallibility has led to multiple nuclear risks in the past, but ironically it most of the time human beings, with all of their faults are the ultimate safeguard in preventing nuclear disaster. One cannot outsource the management of global nuclear dangers to anyone or anything, let alone unproven and unreliable large language models. The only protection against nuclear use is effective and accountable human control and the eventual elimination of all nuclear weapons under strict and effective verification.”
