[Panel Discussion] Artificial Intelligence and Weapons of Mass Destruction

Artificial Intelligence (AI) presents several significant risks to arms control, primarily due to its potential to disrupt the balance of power in geo-politics, introduce unpredictability into decision-making and undermine existing arms control agreements. Existing arms control treaties do not adequately cover AI technologies. Many treaties focus on traditional forms of military hardware, such as nuclear weapons and ballistic missiles, while AI produces new, unconventional threats that are harder to regulate.

AI systems can, for example, generate sophisticated misinformation or stimulate realistic weapons, making it harder to verify treaty compliance or detect violations. The complexity of AI systems also means they can behave unpredictably in certain scenarios. In high-stakes situations like nuclear deterrence, any misstep could have catastrophic consequences.

However, AI can also promote arms control by enhancing verification, improving monitoring and supporting decision-making within international institutions and government. Data processing can aid the efficient and timely detection of treaty violations, such as covert weapons production. AI-drive robots and automated systems can help in the safe dismantling of complex weapons systems, including chemical weapons.

The panel discussion examines the risks and benefits of AI in the field of non-proliferation and disarmament of WMD and asks how we can future proof international law to account for the rapid advancement of these technologies.

Panel:

• Michal Onderco (Professor of International Relations, Erasmus University Rotterdam)
• Jerome Larosch (Team Lead CBRN, Dutch Ministry of Foreign Affairs)
• Peter Hotchkiss (Senior Policy Advisor, OPCW)