Rome Declaration for an Unarmed and Disarming Peace

The Rome Declaration for an Unarmed and Disarming Peace is a voluntary international declaration for the ethical governance of AI. It is not legally binding. Read here to learn more about it.

The Rome Declaration for an Unarmed and Disarming Peace, signed in July 2026 by Nobel laureates, Artificial Intelligence (AI) scientists, religious leaders, philosophers, and global policymakers, has reignited the global debate on the ethical governance of Artificial Intelligence.

The Declaration warns against the “convergent peril” that arises from delegating moral, ethical, and life-or-death decisions to AI systems, particularly in nuclear command and control, autonomous weapons, and military decision-making.

The Declaration calls for an international treaty prohibiting AI systems from independently making decisions related to the use of nuclear weapons. It emphasises that human judgment must always remain central to decisions involving war, peace, and human survival.

What is the Rome Declaration for an Unarmed and Disarming Peace?

The Rome Declaration is a global ethical and strategic appeal advocating responsible governance of Artificial Intelligence and renewed commitment towards nuclear disarmament.

• Rather than functioning as a legally binding treaty, it serves as a normative framework encouraging governments, technology companies, scientific communities, and international organisations to establish ethical safeguards against the misuse of AI.
• The Declaration argues that while AI offers tremendous benefits, allowing autonomous systems to make irreversible military decisions could pose an existential threat to humanity.

Philosophical Foundation

The Declaration draws inspiration from Pope Leo XIV’s encyclical Magnifica Humanitas, which emphasises:

• Protection of human dignity
• Preservation of moral responsibility
• Ethical use of technology
• Human-centred scientific progress

Its central message is that technology must remain subordinate to humanity rather than replace human moral judgment.

Current Status

The Rome Declaration is presently:

• A voluntary international declaration
• A moral and ethical guideline
• Not legally binding
• Intended to influence future international treaties on AI governance and autonomous weapons

Core Principles of the Rome Declaration

1. Preventing an AI Arms Race

The Declaration urges countries to prevent:

• AI-enabled military competition
• Autonomous nuclear command systems
• Escalation of technological warfare

2. Responsible Development of AI

AI systems should be developed:

• ethically
• transparently
• safely
• under internationally accepted principles

Developers should publicly disclose ethical safeguards built into AI systems.

3. Responsible Use of AI

AI must not independently perform functions involving:

• nuclear launch decisions
• lethal force
• military targeting
• autonomous warfare

Instead, AI should remain an advisory tool.

4. Responsible Governance

The Declaration calls for:

• international oversight
• regulatory institutions
• accountability mechanisms
• transparent governance

5. Responsible Leadership

Above technological competition, governments, corporations, scientists, and military establishments must place:

• ethics
• humanity
• global peace

6. Nuclear Disarmament

• The Declaration also renews the global commitment towards the complete, irreversible, and verifiable elimination of nuclear weapons.

Ethical Mandates of the Declaration

Mandatory Meaningful Human Control

The Declaration categorically opposes allowing AI systems to independently decide on:

• launching nuclear weapons
• initiating armed conflict
• employing lethal force

Final authority must always rest with human decision-makers.

Digital Commons

The Declaration promotes a “Digital Commons” model where:

• AI datasets
• safety research
• assessments

These risks are shared among governments, scientists, and independent experts to reduce systemic risks.

Responsible AI Development

Developers must ensure:

• explainability
• auditability
• transparency
• human override mechanisms

AI systems that cannot be controlled or shut down by humans should not be deployed.

Nuclear Infrastructure Security

To prevent AI-enabled cyber attacks or accidental launches, countries possessing nuclear weapons should conduct:

• cybersecurity audits
• AI vulnerability assessments
• command-and-control reviews

Time-Bound Nuclear Disarmament

• The Declaration encourages renewed international negotiations towards gradual, verifiable, and irreversible nuclear disarmament.

Why was the Declaration Needed?

Shrinking Decision-Making Time

Traditional nuclear deterrence allowed leaders time to:

• verify intelligence
• consult advisers
• avoid accidental escalation

Historical examples include:

• Cuban Missile Crisis (1962)
• Soviet False Alarm Incident (1983)

AI-powered military systems may compress decision-making from hours to mere seconds, leaving little room for human judgment.

AI Failure Risks

Modern AI systems remain susceptible to:

• hallucinations
• faulty training data
• adversarial attacks
• cyber manipulation
• algorithmic errors

In military settings, these failures could trigger catastrophic consequences.

Weakening Arms Control

Several global arms control mechanisms have weakened in recent years, including:

• Open Skies Treaty
• New START Treaty

This creates incentives for countries to increasingly automate military systems.

Emerging Geopolitical Flashpoints

Highly volatile regions increase the likelihood that AI-assisted military systems could misinterpret signals and unintentionally escalate conflicts.

• nuclear programmes
• border disputes
• cyber warfare

What are “Convergent Perils”?

The Declaration introduces the concept of Convergent Peril.

Convergent peril refers to the interaction of multiple technological risks that reinforce each other, creating a systemic threat much greater than the sum of individual risks.

In AI, these include:

• algorithmic bias
• lack of accountability
• hyper-automation
• cybersecurity vulnerabilities
• autonomous decision-making
• ethical failures

Together, these may produce large-scale societal or geopolitical crises.

Concerns Over Outsourcing Ethics to AI

Loss of Human Moral Agency

Ethical reasoning requires:

• empathy
• compassion
• responsibility
• contextual judgment

AI cannot genuinely understand human suffering or moral dilemmas.

Delegating ethical decisions to algorithms may weaken humanity’s own moral capacities.

Value Alignment Problem

AI systems often optimise mathematical objectives.

However, human ethics balances:

• justice
• dignity
• rights
• fairness
• compassion

An AI focused solely on maximising overall benefit may disregard individual rights.

Algorithmic Bias

AI learns from historical data.

If datasets contain biases related to:

• gender
• caste
• race
• religion
• socioeconomic status

AI may perpetuate and even amplify discrimination.

Accountability Gap

Advanced AI models often function as “black boxes.”

It becomes difficult to determine responsibility if AI is used wrongly:

• denies healthcare,
• recommends lethal military action,
• misidentifies civilians,

Questions arise regarding the accountability of:

• developers
• governments
• operators
• manufacturers

Lack of Contextual Judgment

Humans consider:

• mercy
• exceptional circumstances
• intent
• compassion

AI lacks emotional intelligence and cannot appreciate the nuances of complex human situations.

Global AI Ethics Initiatives

Several international frameworks already seek to promote responsible AI.

UNESCO Recommendation on the Ethics of AI (2021)

• Human rights
• Transparency
• Accountability
• Sustainability
• Human dignity

EU Artificial Intelligence Act (2024)

• World’s first comprehensive AI legislation based on a risk-based regulatory approach, prohibiting unacceptable AI applications while imposing strict obligations on high-risk systems.

Bletchley Park Declaration (2023)

Signed by 28 countries to promote:

• frontier AI safety
• international cooperation
• responsible AI governance

Global Partnership on Artificial Intelligence (GPAI)

India is a founding member.

The initiative promotes:

• inclusive AI
• responsible innovation
• human-centric AI development

India’s AI Governance Framework

National Strategy for AI (2018)

Prepared by NITI Aayog under the vision of “AI for All.”

It emphasizes:

• inclusiveness
• transparency
• fairness
• privacy
• accountability

IndiaAI Mission (2024)

The mission seeks to strengthen India’s AI ecosystem through:

• computing infrastructure
• indigenous datasets
• startup support
• AI skilling
• innovation
• trusted AI

MeitY AI Governance Guidelines

The Ministry of Electronics and Information Technology, through its Seven Sutras for AI Governance, advocates:

• principle-based regulation
• trustworthy AI
• safe AI
• explainable AI
• responsible innovation

Way Forward

Human-in-the-Loop (HITL)

Critical decisions involving:

• healthcare
• military operations
• criminal justice
• nuclear weapons

must always require meaningful human oversight.

Explainable AI

Governments should mandate:

• transparency
• explainability
• auditability

particularly for high-risk AI applications.

Ethical Impact Assessment (EIA)

Before deployment, AI systems should undergo independent ethical assessments similar to environmental impact assessments to evaluate:

• fairness
• bias
• human rights implications
• societal risks

AI by Design

Ethical principles should be embedded into AI systems from the design stage itself.

• fairness
• accountability
• privacy
• human dignity

Global Treaty on Autonomous Weapons

• The international community should negotiate a legally binding treaty prohibiting Lethal Autonomous Weapon Systems (LAWS) that operate without meaningful human control.

Conclusion

The Rome Declaration for an Unarmed and Disarming Peace highlights one of the defining ethical challenges of the twenty-first century: ensuring that Artificial Intelligence remains a tool that serves humanity rather than replaces human moral judgment.

As AI increasingly shapes military strategy, governance, healthcare, and public decision-making, preserving meaningful human control, transparency, accountability, and ethical responsibility becomes indispensable.

Strengthening global cooperation through frameworks such as the UNESCO AI Ethics Recommendation, the EU AI Act, and India’s IndiaAI Mission will be critical to ensuring that technological progress advances peace, human dignity, and global security rather than creating new existential risks.