As geopolitical tensions escalate in Europe, the Middle East, and Asia, new technologies such as Artificial Intelligence (AI) I are developing at a rapid rate. Its adoption by nuclear armed states, say experts, further impacts strategic stability and complicates the landscape of nuclear risk. 

Against this backdrop, the Normandy P5 Initiative has served as a powerful platform working toward global nuclear risk reduction. The Initiative, co-convened by the Strategic Foresight Group and the Geneva Centre for Security Policy, draws inspiration from the Normandy Manifesto for World Peace, which was issued in 2019 by a group of Nobel Peace Laureates and other social thinkers. The Manifesto re-emphasized the existential risk to humanity that nuclear war poses and warned that a global nuclear war could take place due to intent, incident, or accident. If present or future leaders of nuclear powers intend to wage a nuclear war, human civilization risks becoming extinct. And it is in no one’s interest to trigger a nuclear war by incident or accident.

The aim of these dialogues is to bolster the commitment to the 1968 Nuclear Non-Proliferation Treaty (NPT) – a binding treaty to prevent the spread of nuclear weapons

The Normandy P5 Initiative seeks to engage the five permanent members of the UN Security Council (China, France, the Russian Federation, the United Kingdom and the United States) in finding solutions to mitigate or eliminate nuclear risk through a “track two” dialogue of strategic experts. This dialogue is critical for minimizing the risks of misunderstanding and miscalculation. To date, this dialogue has been conducted through a series of roundtables, the most recent of which occurred in Geneva in December of 2023. In addition, the convenors regularly hold bilateral consultations with the Disarmament Ambassadors of the P5 countries to keep “track one,” or formal negotiations, informed.

The experts of the roundtable have stressed the continued importance and urgency of nuclear risk reduction, particularly considering the escalated risks presented by emerging technologies and geopolitical destabilization. Although no consensus document was issued in Geneva, the co-convener’s summary presented a set of key observations and recommendations that can guide the future of this dialogue, along both formal and informal tracks. 

The aim of these dialogues is to bolster the commitment to the 1968 Nuclear Non-Proliferation Treaty (NPT) – a binding treaty to prevent the spread of nuclear weapons – and to generate momentum for a more meaningful NPT review conference in 2026. A series of working level meetings held to discuss doctrines, risk reduction strategies, and other NPT-related issues over the past year have proven beneficial. While the role of the P5 is crucial, greater outreach and involvement of other parties to the NPT, and other nuclear-armed states, is also essential.

Other recommendations made by the group toward nuclear risk reduction include:

1. Embrace Joint Statements: Stress the shared responsibilities and mutual interests of the P5 by building on their 3 January 2022 Joint Statement on Preventing Nuclear War. Other possibilities for similar statements include upholding the Comprehensive Test Ban Treaty, which would commit each state to confirm that they would not be the first to start testing again.
2. Include Space Security on the Agenda: Add space security, in the context of the NPT, to the agenda of the official “P5 Process.” Specifically, the P5 should declare their threat perceptions to each other and explain their space policies/doctrines. 
3. Include AI on the Agenda: Add the AI-nuclear weapon nexus to the agenda of the official ‘P5 Process,’ initially at the working level.
4. Expand the Scope of the Nuclear Risk Reduction Vision: Expand the transparency of each country’s threat perceptions regarding nuclear risk, space security, and artificial intelligence; develop shared understandings on topics important to strategic stability, such as ballistic missile defense, hypersonic missiles, and delivery systems; ensure transparency regarding nuclear testing activities; and expand dialogue on nuclear weapons and missiles. 
5. Expand Non-Proliferation Dialogues: Ensure the creation of formal dialogues on non-proliferation issues in the Middle East and Korean Peninsula. 
6. Support Verification Efforts: Encourage the P5 to continue their constructive engagement on nuclear disarmament verification.
7. Engage the Next Generation of Experts: Continue support for the next generation of nuclear experts through initiatives such as the Young Professionals Network. 

Additionally, the P-5 experts held a discussion on the nexus between AI and nuclear command, control, and communications (NC3). They stressed that the P5 needed to strive for a shared understanding of the impact of AI on the nuclear domain and be sensitive to the risks it posed. The latter include over-confidence in its readings, manipulation of early warning systems, multiple AI systems working against each other, and unclear responses to unpredictable risks. 

From that discussion came an additional set of recommendations specific to the AI-nuclear nexus:

1. Affirm Human Control of NC3: Issue a joint declaration that humans will always make decisions related to NC3, a valuable public reassurance measure. Such a declaration could be endorsed by the UN Security Council. 
2. Support Cross-Disciplinary Collaboration and Engagement: Support more informal dialogue initiatives that aim to bring together AI experts, computer scientists, and the Big Tech sector with experts on nuclear weapons. In this way, the implications of the increasing use of AI in NC3 can be better understood by all stakeholders. 
3. Expand AI-Specific International Dialogue: States should clarify how the increase in use of AI will affect their nuclear doctrines, discuss their understanding of AI terminology, and share best practices on national safety reviews for using AI in NC3. 
4. Make AI a Central Focus of the 2026 NPT Review Conference: Support a dedicated Thematic Session at the next NPT Review Conference on the AI-nuclear nexus. 
5. Normalizing National Determination of Red Lines: Nationally, each of the P5 should determine their behavioral red lines for AI, with red teaming exercises to work out how they would respond in certain scenarios. 

The Geneva roundtable advances the nuclear risk reduction effort, while underscoring the broad commitment of states and global experts to address this existential risk. In addition, the roundtable’s inclusion of emerging technology and AI-related factors that could increase the risk of nuclear catastrophe put this issue in a fresh, but also more urgent, light. The group’s recommendations, focused on dialogue and transparency, seek to chart a safer way forward for the global community. While the Normandy P5 Initiative will continue to engage the P5 members, the rest of the world cannot ignore growing nuclear risks. Once a bomb is detonated by a military commander, or by an algorithm, the survival of humanity will be endangered. Hence, it is also necessary for the international community at large to demand accountability from the P5 and other nuclear powers.