Global Figures Unite Over AI and Nuclear Risks
More than 200 participants gathered in Italy from July 14 to 16, 2026, for the Global Nobel Laureates Assembly on Artificial Intelligence and Nuclear War.
The first two days were hosted at Borgo Laudato Si’ in Castel Gandolfo, while the concluding session took place at Rome’s Capitoline Hill. The gathering ended with the signing of the Rome Declaration for an Unarmed and Disarming Peace in the Age of Artificial Intelligence, Nuclear and Autonomous Weapons, New Digital Protocols, and Emerging Models of Digital Development.
Participants included Nobel laureates, researchers, diplomats, religious representatives, peace advocates and former heads of state and government.
Their central message was that technological systems may assist human decision-making, but they must not become the final moral authority over matters of life, death, peace or war.
Human Responsibility Must Remain Central
Machines Cannot Carry Moral Accountability
The declaration argues that humanity must not transfer responsibility for existential decisions to algorithms or autonomous systems.
Cardinal Baldo Reina, the Vicar General of Rome, said no machine or algorithm should be placed at the centre of decisions on which humanity’s survival depends. He stressed that choices concerning war, peace and future generations must remain subject to responsible and meaningful human control.
This principle addresses a fundamental limitation of AI: a machine can process information and optimise toward programmed objectives, but it does not bear moral responsibility for the consequences.
Political leaders, military commanders, scientists and technology developers therefore remain accountable for how such systems are designed and used.
Why AI and Nuclear Weapons Are Converging Risks
Automation Could Compress Decision Times
Artificial intelligence is increasingly being incorporated into intelligence analysis, surveillance, targeting, cyber operations and military planning.
These systems may help governments process enormous amounts of information more quickly. However, speed can also become a danger during a crisis.
In a nuclear confrontation, automated warnings or AI-generated assessments could reduce the time available for leaders to verify information, challenge assumptions or pursue diplomacy.
A technical error, manipulated data stream or mistaken interpretation could create pressure for an immediate response before human decision-makers fully understand what has happened.
The Rome gathering therefore treated AI risk and nuclear danger as connected problems rather than separate policy debates.
Fear of Autonomous Nuclear Decision-Making
The declaration’s supporters are particularly concerned about systems that could influence or automate nuclear command-and-control processes.
Nuclear decisions already take place under extreme time pressure. Introducing increasingly autonomous technology could weaken the hesitation, judgement and moral restraint that human beings may exercise before authorising irreversible action.
Participants argued that AI should never be given independent authority to initiate lethal force, especially where nuclear weapons are involved.
They also called for stronger international cooperation and measures by nuclear-armed countries to reduce the risk of accidental or deliberate nuclear conflict.
Nobel Laureates Warn of a Renewed Arms Race
David Gross Says the Danger Has Increased
Nobel Prize-winning physicist David Gross warned that the nuclear threat is more serious than it was several decades ago.
He pointed to the erosion of arms-control agreements, the existence of nine nuclear-armed states and the acceleration of military competition.
Gross urged nuclear powers to adopt policies that reduce the possibility of nuclear war and annihilation. He also reminded participants that the consequences would extend beyond the current generation to children and grandchildren.
His warning reflects growing concern that AI could become another layer in an already unstable arms race, with countries feeling compelled to deploy increasingly autonomous systems because rivals are doing the same.
Maria Ressa and Other Leaders Call for a Moral Compass
Nobel Peace Prize laureate Maria Ressa was among the prominent figures who addressed the gathering.
Speakers emphasised that rapid scientific progress must be accompanied by an equally serious moral and political response.
Peace Ambassador Sharon Stone argued that as machines become more capable, the ethical efforts of their creators must also grow. Her intervention reinforced the declaration’s broader position that human dignity cannot be reduced to a mathematical calculation or automated output.
AI Offers Benefits as Well as Risks
The declaration does not reject artificial intelligence as inherently harmful.
Participants acknowledged that AI could support major advances in:
Healthcare and disease detection
Education and scientific research
Environmental monitoring
Public health
Poverty reduction
Conflict prevention and peacebuilding
The concern is not simply the existence of powerful technology, but how it is governed and the objectives it is asked to serve.
AI used to analyse climate data or improve medical diagnosis presents a very different ethical challenge from AI used to select military targets or accelerate decisions during a nuclear crisis.
Background: Technology Is Moving Faster Than Governance
Artificial intelligence has developed faster than many governments’ ability to create comprehensive rules for its deployment.
At the same time, international arms-control structures have weakened, and military powers are investing heavily in autonomous systems, cyber capabilities and AI-supported intelligence.
This creates a dangerous combination: rapidly advancing technology, geopolitical rivalry and insufficient international agreement.
The Rome Declaration seeks to address that gap by connecting AI governance with nuclear disarmament, peacebuilding and the protection of human dignity.
What Meaningful Human Control Should Involve
The phrase “meaningful human control” is often used in discussions about autonomous weapons, but its practical definition remains contested.
For oversight to be meaningful, a person must have more than the ability to approve a machine’s recommendation automatically.
Human decision-makers should:
Understand the system’s limitations
Have sufficient time to review its conclusions
Be able to reject or override recommendations
Remain legally and morally accountable
Receive reliable information about uncertainty and possible errors
Avoid treating automated outputs as unquestionable facts
Without these safeguards, human involvement risks becoming symbolic rather than real.
Why Human Hesitation Can Be a Safeguard
In many technological contexts, hesitation is considered inefficient.
In decisions involving nuclear weapons or lethal force, however, hesitation may prevent catastrophe.
A human being can recognise uncertainty, question an order, consider the suffering involved and choose not to act. An automated system may instead follow its objective at extraordinary speed without experiencing fear, compassion or moral doubt.
The declaration’s supporters therefore view human judgement not as an obstacle to efficiency, but as an essential safety mechanism.
The Challenge of Accountability
Who Is Responsible When an AI System Fails?
AI-assisted military decisions may involve software developers, data providers, defence contractors, commanders and political authorities.
When something goes wrong, responsibility can become fragmented.
Officials may blame incorrect data, developers may blame misuse, and institutions may argue that the algorithm behaved unexpectedly.
The Rome Declaration’s emphasis on human control is intended to prevent this form of moral outsourcing.
AI may provide advice, but responsibility must remain traceable to identifiable people and institutions.
Calls for Stronger International Cooperation
The declaration urges governments to rebuild international cooperation around nuclear risk reduction and emerging technologies.
No country can manage these dangers entirely on its own because AI systems, cyber operations and nuclear threats cross national borders.
Possible areas for cooperation include:
Rules preventing autonomous nuclear-launch authority
Transparency about military AI systems
Crisis communication between nuclear powers
Testing and verification standards
Protection against manipulated warning data
Agreements governing autonomous weapons
Renewed nuclear arms-control negotiations
The declaration itself is not a binding treaty, but its supporters hope it can influence governments and help shape future international agreements.
Why This Story Matters
The debate is no longer limited to whether AI can write documents, generate images or improve office productivity.
Artificial intelligence is increasingly entering areas where errors can affect national security and human life.
If governments allow machines to shape military decisions without adequate oversight, technological speed could override diplomacy, proportionality and moral judgement.
The declaration matters because it reframes AI governance as a question of peace, responsibility and human survival—not merely innovation or economic competition.
Balanced Analysis
The Rome Declaration delivers a powerful ethical warning, but translating its principles into enforceable policy will be difficult.
Governments may agree publicly that humans should remain in control while continuing to develop increasingly automated military systems.
There is also no universally accepted definition of meaningful human oversight. A military may claim that a person approved an operation even when that person had only seconds to review an opaque algorithmic recommendation.
At the same time, banning all military uses of AI could be unrealistic. AI may help detect threats, reduce civilian casualties or improve defensive systems when used carefully.
The central policy challenge is therefore not choosing between total adoption and total rejection. It is determining where automation becomes morally unacceptable and creating rules that prevent governments from crossing that line.
Can Declarations Produce Real Change?
International declarations do not automatically alter national policy.
Their influence depends on whether political leaders, military institutions, technology companies and the public adopt their principles.
However, such statements can establish ethical norms, guide negotiations and create pressure for binding agreements.
The Rome Declaration may be most significant as an effort to bring scientific, religious and political voices into the same conversation.
Its supporters are arguing that AI governance cannot be left solely to engineers, corporations or defence agencies because its consequences affect society as a whole.
Conclusion
The Rome Declaration presents a clear warning: humanity must not outsource its conscience to artificial intelligence.
AI can support analysis and improve decision-making, but machines should not determine when lives are taken, wars begin or nuclear weapons are launched.
As artificial intelligence and military technology advance together, the demand for responsible human control is becoming increasingly urgent.
The declaration’s long-term impact will depend on whether governments move beyond ethical statements and create enforceable international safeguards before technological competition makes restraint even harder.
